Re: What to do with d-i on armel?
On 09/01/2024 23:33, gene heskett wrote: 3d printing environment relies heavily on whatever runs the rp2040. A whole industry has grown up around the combo of the adxl345 and an rp2040 to measure resonances and tune them out, allowing the printer to run several times faster with what is called "input shaping" The actual device is about the size of a postage stamp on a teeny pcb. I do not know which version of linux it is that the rp2040 is actually running. For us, its one of those things that Just Works. The rp2040 does not run linux.
Re: Raspberry Pi 5
On 28/11/2023 17:52, Robert Wilkinson wrote: Hello What chance do I have of getting Debian to run on my new RPi 5 ? Pure Debian will likely be possible in the long term, but it may take a while. The Debian kernel team refuse to package vendor downstream kernels (ubuntu are more flexible on this). So the first thing that needs to happen is adequate support for the Pi 5 in upstream kernels. Some searches turned up that there does seem to be some raspberry pi 5 related activity on the upstream kernel mailing lists (e.g. https://www.spinics.net/lists/kernel/msg4986331.html ) Once there is adequate support in upstream kernels, then the Debian packaging can be updated to support it. In the meantime it's almost certainly possible to run a pure Debian userland with a kernel from Raspberry pi.
Re: Just tried arm64 netinstall on a bananai-m5
On 15/08/2023 17:44, gene heskett wrote: used dd to write the arm64-bookworm-12.1 netinstall image to a 64G SDXC ONN. brand card, makes no attempt to boot plugged into a bananapi-m5. bring card back to reader, can't mount it, wrong filesystem for both partitions. Give up, write Armbian-jammie-full-desktop iso to card, mounts ok, boots bananapi-m5 normally. What did I do wrong? The unfortunate reality is that boot on arm is *still* a mess. The server guys and the windows laptop guys have settled on uefi (though the implementations are often far from perfect), but the hobbyist board segment is still all over the place, with each board (or family of closely related boards) still needing it's own build of u-boot that knows how to initialise the board, load a kernel and initrd and pass them the relavent device tree. For some boards, Debian offers "concatenatable images", where a board-specific boot section can be concatenated with a board-independent d-i section to produce a boot image suitable for a specific board, yours doesn't seem to be one of them though. .
Re: buildd reliability
On 30/03/2023 19:59, Wookey wrote: OK. I'll see what can be done. I see Altra servers are from $7000-$53000 on https://store.avantek.co.uk/arm-servers.html. What does DSA consider 'decent'? I guess we'd prefer the resilience of a couple of reasonable machines over one ridiculously manly one. A bit of configury on the Aventek site suggests that basic ARM Altra servers cost about twice as much as AMD ones for similar specs (cores/RAM/disk), but then the power consumption is less than half. I don't know how the performance actually compares for buildd purposes (nor what sort of spec we prefer in terms of nodes/cores/RAM/Disk/networkIF), but people describe the Altra's as 'fast'. I'll try and collect some more details to quantify that. I'm not DSA, but the problem i've found when looking at such hardware has always been that the "cost of entry" is so high. To me $7K is a lot of money to put down "on spec". It may well be worth approaching them to see if they would sponsor some hardware, or at least provide a steep discount from list prices. Debian is after-all one of the top Linux distros (and probablly the top "community" distro). Also can anyone confirm if these machines support arm32? or if they are arm64 only?
Re: Bug#1001314: mozjs78: FTBFS on armhf: '-mfloat-abi=hard': selected architecture lacks an FPU
On 08/12/2021 23:56, Simon McVittie wrote: > On Wed, 08 Dec 2021 at 20:11:55 +0000, peter green wrote: >> The default -march value on Debian armhf is "armv7-a+fp". You should >> *NOT* use "armv7-a+vfpv3" as that specifies the version of vfpv3 >> with 32 double precision registers which is above the Debian baseline. > > https://github.com/alexcrichton/cc-rs/commit/b2f6b146b75299c444e05bbde50d03705c7c4b6e > (which I have now applied to the copy of cc-rs in mozjs78) uses > "-march=armv7-a -mfpu=vfpv3-d16", which seems like a plausible encoding of > the minimum spec given in https://wiki.debian.org/ArmHardFloatPort and > https://wiki.debian.org/ArchitectureSpecificsMemo#Architecture_baselines > (ARMv7 with Thumb-2 and VFPv3-D16). Yes that is fine. > > If I understand correctly, VFPv3-D16 refers to the version of vfpv3 with > only 16 registers, which is exactly what our armhf baseline guarantees > (and as you point out, some ARM CPUs with VFPv3 have more registers, but we > require our binaries to assume those extra registers are not available). > Am I getting this right? Yes.
Re: Bug#1001314: mozjs78: FTBFS on armhf: '-mfloat-abi=hard': selected architecture lacks an FPU
On 08/12/2021 09:48, Simon McVittie wrote: On Wed, 08 Dec 2021 at 09:41:29 +, Simon McVittie wrote: At a guess, perhaps the problem is that the mozjs build system is explicitly specifying -march=armv7-a when it should be something like -march=armv7-a+vfpv3 or accepting the compiler's default? Looks like this might be the same issue fixed by https://github.com/alexcrichton/cc-rs/commit/b2f6b146b75299c444e05bbde50d03705c7c4b6e Certainly looks like it to me. The Debian gcc packages changed how they specify floating point options, they now use "-march" to set both the CPU and FPU rather than using seperate "-march" and "-mfpu" options. The result of this change is if you manually set "-march" then it overrides the built-in defaults for both the CPU and FPU rather than only overriding the CPU. If your new "-march" setting doesn't specify a FPU and you don't set a seperate "-mpfpu" option then you end up with an invalid configuration (hard float ABI selected but no FPU selected). The default -march value on Debian armhf is "armv7-a+fp". You should *NOT* use "armv7-a+vfpv3" as that specifies the version of vfpv3 with 32 double precision registers which is above the Debian baseline. https://gcc.gnu.org/onlinedocs/gcc/ARM-Options.html
Re: Help with boost1.74
On 09/11/2021 19:51, Giovanni Mascellani wrote: Dear Debian ARM people, last boost1.74 version (1.74.0-12) fails to build on armhf[1], and I suspect the failure is caused by the switch to GCC 11, because it didn't happen in the version before (compiled with GCC 10). IIRC the Debian gcc-11 package takes a different approach to specifying default cpu/fpu, specifying -march=armv7-a+fp rather than seperate march and mfpu options, I wonder if this is having an influence somehow.
Re: W: armhf version for thunderbird bullseye
I think the issue may have been the linker running out of address space. Certainly we ran into that in raspbian. I do have a thunderbird package in raspbian which may work for you, but I have to build it in a slightly hacked-up environment with the linker replaced by a cross-linker. On 06/10/2021 16:06, John Paul Adrian Glaubitz wrote: Hello! On 10/6/21 16:38, Tuxo wrote: Is there a thunderbird version for armhf in one of the debian repositories? checking [1] and [2] they mention armhf versions of tb 78 for Buster and Stretch, none for Bullseye so far. Carsten Schoenert has disabled the 32-bit ARM builds of Thunderbird, see [1]. I have no clue why that happened as there is no mention of that change in debian/changelog, but I assume it's because Mozilla upstream broke support on these targets. It's unfortunately a big of cat and mouse with fixing all these regular regressions on architectures that Mozilla doesn't care about. We would need more people working on these. Adrian [1] https://buildd.debian.org/status/package.php?p=thunderbird=sid
Re: armhf SIGILL, Illegal Instruction
On 29/09/2021 23:39, Jeffrey Walton wrote: On Wed, Sep 29, 2021 at 5:05 PM peter green wrote: As I understand it, there are two variants of "VFPv3", a version with 32 double registers (d0 to d31) and a version with only 16 double registers (d0 to d16). The former is reffered to by gcc as "vfpv3" while the latter is reffered to by gcc as "vfpv3_d16". Debian is supposed to support vfpv3_d16 but because there is relatively little hardware out there that doesn't support the extra registers bugs may take a while to get noticed. So IMO this is a bug in the compiler that is generating that code. What i'm not so sure about is whether selecting the correct compilation settings is the responsibility of the frontend (ldc) or the backend (llvm). Shouldn't that show up in the build logs? It will only show up in build logs if the build process is overriding the built-in defaults of the compiler. Normal practice in Debian is that when invoked without specific architecture flags compilers should generate code that will run on the baseline CPU of the port. If they don't then that is a bug in the compiler.
Re: armhf SIGILL, Illegal Instruction
As I understand it, there are two variants of "VFPv3", a version with 32 double registers (d0 to d31) and a version with only 16 double registers (d0 to d16). The former is reffered to by gcc as "vfpv3" while the latter is reffered to by gcc as "vfpv3_d16". Debian is supposed to support vfpv3_d16 but because there is relatively little hardware out there that doesn't support the extra registers bugs may take a while to get noticed. So IMO this is a bug in the compiler that is generating that code. What i'm not so sure about is whether selecting the correct compilation settings is the responsibility of the frontend (ldc) or the backend (llvm). On 29/09/2021 21:06, Ash Hughes wrote: Hi, I've been getting some programs terminated with SIGILL today, and I'm trying to find out if this is a package issue or if Debian (Bullseye) is no longer compatible with my ARM machine. I first got an error with onedrive, with gdb output: #0 0xb6948ca8 in gc.impl.conservative.gc.Gcx.fullcollect(bool) () from /usr/lib/arm-linux-gnueabihf/libdruntime-ldc-shared.so.94 which is "vldr d18, [pc, #216] ;". I then tried to run ldc2, and I got something similar: Core was generated by `ldc2 -c --output-o -conf= -w -mattr=-neon -O3 -release -relocation-model=pic -d'. Program terminated with signal SIGILL, Illegal instruction. #0 0x0089e15c in dmd.parse.Parser!(dmd.astcodegen.ASTCodegen).Parser.parsePrimaryExp() () which is also a vldr instruction ("vldr d16, [r6, #80] ; 0x50") Finally, I tried to compile ldc2 myself and running it I got: #0 0xb4a6eabc in ?? () from /usr/lib/arm-linux-gnueabihf/libLLVM-11.so.1 also vldr ("vldr d16, [sp, #8]") It looks like the vldr instruction is being used in several LLVM packages, in a way my CPU doesn't like. Here's my cpuinfo: processor : 0 model name : ARMv7 Processor rev 1 (v7l) BogoMIPS : 37.39 Features : half thumb fastmult vfp edsp thumbee vfpv3 vfpv3d16 tls idivt CPU implementer : 0x56 CPU architecture: 7 CPU variant : 0x1 CPU part : 0x581 CPU revision : 1 Hardware : Marvell Armada 370/XP (Device Tree) Revision : Serial : I don't have neon, although I think armhf doesn't require it, unless this has changed for Bullseye? If neon isn't required for Debian armhf, does this mean some LLVM related packages could be built differently to improve compatibility? Thanks, Ash
Re: Replacement for raspPi
On 07/09/2021 11:32, Keith Bainbridge wrote: G'day I've been following the recent thread Subject: Re: Debian on Pine64 H64B? I'm looking for suggestions for a new SBC, please. Ideally something more than 2M RAM. I see that a few a happy with Pine laptops. Does this translate to their SBC? Thanks Background. I've just set up a new SDCard with raspberryOS. Set myself up as a user, and set root passsword. When I sshfs to my laptop, the mount point AND all subdirectories are owned exclusively buy user pi. Not even root has access. So you had a problem with sshfs permissions, rather than investigate said problem (e.g. https://unix.stackexchange.com/questions/59685/sshfs-mount-sudo-gets-permission-denied ) you chose to blame the Raspberry Pi.
Re: Porter roll call for Debian Bullseye
On 08/12/2020 21:02, Lennart Sorensen wrote: > My understanding is that Debian never cross compiles Indeed Debian always builds native, though "native" does include running 32-bit userlands on 64-bit kernels. > And some packages would take way to long to compile on small machines > like a pi4 due to not enough ram. Pi4s are now available with 8GB of ram, not as much as the "server" hardware granted, but enough to pretty much max out the address space for two 32-bit processes at the same time. > And for build machines, you want rack > mountable machines You can rackmount damn near anything if you put a bit of effort in. > with remote management. Not something that doesn't > even have a power switch or ability to be restarted remotely if needed. A Pi4 can be set up with a serial console and netboot and turns on when the power is applied (no need to press any buttons). Contrast this to the asus arm64 server board I have access to, which according to the hoster needs to be poked over IPMI after a power cycle before it will turn back on and which sometimes fails to come back after a soft reboot. In terms of power cycling, what Mythic Beasts do for their hosted Pis is to use a POE hat, then turn on and off the port power on the network switch. They don't have HDDs/SSDs attached though, which you would want for a buildbox, and i'm not sure if the PoE hat as the current capability to reliably run a Pi4+HDD.
Re: Porter roll call for Debian Bullseye
On 08/12/2020 16:40, Lennart Sorensen wrote: On Tue, Dec 08, 2020 at 09:21:42AM +0100, basti wrote: Hello Adrian, How can I help to get bullseye on armel? My impression was that the biggest problem for armel and armhf is the lack of reliable build machines. Not sure if any of the 64 bit arm servers are suitable for building armel or armhf or if they lack certain required instructions. Debian has been building armel and armhf on arm64 "server" hardware for some time. There have been some issues relating to building on 64-bit kernels (mostly that the 64-bit kernels default to trapping alignment errors rather than fixing them up), but nothing earth-shatteringly bad. You do of course have to make sure you get a machine that actually has 32-bit support, some of the arm server hardware doesn't.
Re: Bug#976567: libhmsbeagle: FTBFS: GPUInterface.h:221:27: error: there are no arguments to ‘malloc’ that depend on a template parameter, so a declaration of ‘malloc’ must be available [-fpermissive]
On 05/12/2020 20:26, Andreas Tille wrote: Control: tags -1 help Hi, I need to admit that I have no idea why this error occures on arm64. According to reproducible builds, it's also happening on i386 and armhf, it's showing a pass for amd64 but that could just be because it hasn't been tested for a while (I've just hit the re-test button) Any hint would be welcome Looks like you need to add #include to libhmsbeagle/GPU/GPUInterface.h I would guess stdlib.h was previously pulled in implicitly by some other header and no longer is.
re: bio-eagle: FTBFS: MemoryUtils.hpp:34:10: fatal error: xmmintrin.h: No such file or directory
severity 976572 normal thanks During a rebuild of all packages in sid, your package failed to build on arm64 (I don't know if it also fails on amd64). It's cool that you have expanded your rebuild tests to include arm64, but it seems your test workflow needs some work. arm64 is not on the architecture list for bio-eagle, even if it was, the normal practice in Debian is only to consider a FTBFS bug rc if the package has built on the architecture in the past. I have downgraded the bug, I will leave the decision on whether to close it or not up to the maintainer.
Re: eight-core ARM64 networking platform with mainline Linux support
On 14/11/2020 14:47, Geert Stappers wrote: Hi, ARM64 hardware about to hit the market Traverse Ten64 an eight-core ARM64 networking platform with mainline Linux support More information at https://www.crowdsupply.com/traverse-technologies/ten64 Regards Geert Stappers Did order one and would like to known about simular products The clearfog/honeycomb from solidrun springs to mind.
Re: Bug#970132: buster-pu: package rustc/1.41.1+dfsg1-1~deb10u1
> > I'll take a look, can't promise anything but I've had to deal with similar issues > in raspbian before. No, my idea (treat it like a cross-build) didn't work.
Re: Bug#970132: buster-pu: package rustc/1.41.1+dfsg1-1~deb10u1
On 17/09/2020 15:06, Emilio Pozuelo Monfort wrote: On 12/09/2020 11:09, Emilio Pozuelo Monfort wrote: This updates buster's rustc to 1.41, as needed by the new firefox 78 ESR. The bootstrap happens with the upstream binaries as we've done in the past. I have also avoided the bump to LLVM 9/10, we use buster's LLVM 7 instead. The updated rustc is in buster-proposed-updates now, but it failed to build for armel. Looks like my change to debian/architecture.mk to use arm-unknown-linux-gnueabi rather than armv5te-unknown-linux-gnueabi broke this a bit. The reason for that change was that there were no armv5te binaries upstream, and the arm ones are ARMv5 so should be good to bootstrap rustc on armel. However, rustc now defaults to ARMv6 for arm-unknown-linux-gnueabi (see vendor/cc/src/lib.rs), which could be causing some issues. Or perhaps the issue lies elsewhere. Anyway my attempts at getting rustc 1.41 bootstrapped on armel have failed so far (I have tried to lower that -march to armv5te, and also attempted to build for armv5te-unknown-linux-gnueabi while using the arm-unknown-linux-gnueabi binaries). Any help from the armel porters would be appreciated. I'll take a look, can't promise anything but I've had to deal with similar issues in raspbian before.
Re: The state of Arm64 on Raspberry Pi (and its Documentation
On 16/04/2020 19:27, Wookey wrote: Apologies for the confusion. I was rather hoping more projects would use the obvious (and IMHO more user-friendly) arm64 name, rather than following the corporate steer, and in the early days it was hard to tell how this would go. But most have plumped for aarch64, so users are exposed to both names. I think Debian made the right decision in using "arm64", IMO the clarity for end-users (who aren't likely to have a clue what "aarch" is but will probablly have heard of arm) outweighs the inconsistency between Debian and projects that decided to do what arm corporate wanted. Interestingly andriod seems to use arm64 for the "abi"* and aarch64 for the "instruction set" * I am not an andriod expert, but an andiod "abi" seems to be roughly equivilent to a Debian architecture.
Re: Raspberry Pi
On 02/03/2020 19:14, Alan Corey wrote: That poses an interesting question: can you install Debian debs on a Raspbian system and vice versa? In principle Raspbian is just a change of minimum CPU requirements from Debian armhf. So provided your CPU meets the minimum requirements for both, it should be possible to mix raspbian packages with debian armhf packages. This is in-fact how raspbian was initially bootstrapped and I still use it on occasion as a method to rebootstrap stuff. I can't rule out the possibility though that some libraries will change ABIs as a result of the change in minimum CPU requirements, so there is some risk in such mixing. There is also the possibility of multiarching raspbian with another port of Debian. Unfortunately due to the design of multiarch it is necessary to build some extra packages for the architecture you are multiarching with to make this work, I built such packages for arm64 for buster, see https://www.raspberrypi.org/forums/viewtopic.php?t=254027
Re: Armbian
On 03/02/2020 01:20, Paul Wise wrote: The other problem with image-based installation methods is that there are a ridiculous number of devices, so you have to either limit your device support, produce a prohibitively large amount of device-specific images, or figure out how to create one image that works on all devices (which won't be possible due to bootloader diversity on ARM), D-I "solved" this problem by creating "concatenatable images", so you have a small stub image that is board-specific and contains the partition table and first stage bootloader, which you concatenate to the main board-independent image. I don't see any reason why you couldn't do the same for pre-installed images.
Re: SolidRun HoneyComb LX2K
On 16/01/2020 16:35, Phil Endecott wrote: Does anyone have one of these? https://www.solid-run.com/nxp-lx2160a-family/honeycomb-workstation/ I don't have one, but two things strike me 1. The price, £750 basically puts this at a level where you would have to be a total arm fanatic to buy one as a general desktop/server machine (buying one for an arm buildbox may be justifiable to some, so might buying one as a networking role). 2. Support for upstream stuff still seems to be "work in progress", combined with the above high price I would consider the risk of orphaning to be non-negligible. Mini-ITX 16 x Coretex-A72 @ 2 GHz (NXP) 2 x SoDIMM RAM slots 1 x M.2 SSD 1 x PCIe x8 slot 1 x Gigabit ethernet 4 x SFP+ ports 4 x SATA etc. etc.
Re: help with racket build failure
my first suspicions when a debugger can't give a useful backtrace are stack corruption or a call down a bogus function pointer. Unfortunately such problems can be a nightmare to debug, I have often resorted to debugging with printf statements in the past. Recently "reverse debugging" tools have appeared that in principle should be able to help with this sort of thing, but I haven't found any free ones for arm linux. On 13/01/2020 11:36, David Bremner wrote: Dear Arm experts; Any help with https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=948789 would be appreciated. I can replicate the crash on harris, and cd $HOME/racket-7.5+dfsg1/build && \ gdb --args ./racketcgc -cu \ $HOME/racket-7.5+dfsg1/src/racket/src/compile-startup.rkt cstartup.inc \ cstartup.zo $HOME/racket-7.5+dfsg1/src/racket/src/startup.inc \ $HOME/racket-7.5+dfsg1/src/racket/src/schvers.h should get you to the segfault in gdb. The backtrace doesn't make much sense to me, lots of talk about It's entirely possible racket is doing something odd with signals; I remember something like using SEGFAULTS in gc. d
state of d on armhf.
Hi A couple of months ago the default d compiler on armhf was changed from ldc to gdc, the changelog entry was simply "Update list of default compilers forsupported architectures (Closes: #939375)" and bug 939375 makes no mention of armhf. Unfortunately a number of packages have failed to make the transition. appstream-generator: "Error /usr/share/gir-1.0/GObject-2.0.gir(48): Unexpected tag: function-macro in GirPackage: gobject" https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=942379 diet-ng: "gdc: error: unrecognized command line option ‘-main’; did you mean ‘-Wmain’?" no bug currently glib-d: duplicate symbols, no bug currently. gtk-d: duplicate symbols https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=942607 lix: explicitly depends on ldc rather than using the default d compiler, doesn't seem to depend on any d libs, so this may not actually be a problem per-se. onedrive: explicitly depends on ldc rather than using the default d compiler, doesn't seem to depend on any d libs, so this may not actually be a problem per-se. tilix: blocked by gtk-d This leaves me with two questions. 1. Why was the default D compiler on armhf changed? 2. Where do we go from here to restore things to a sane state? (i.e. a state where all d packages in testing can be succesfully rebuilt in testing) Note: I am the maintainer of a derivative and an arm porter, but I don't have any particular knowledge or interest in d.
Re: Debian Installer on Pine64+ (no serial console)
On 07/05/19 07:06, Andrei POPESCU wrote: On Ma, 07 mai 19, 00:33:19, peter green wrote: On 04/05/19 18:49, Andrei POPESCU wrote: https://wiki.debian.org/InstallingDebianOn/PINE64/PINEA64 This prompted me to reinstall my pine64 which has been sitting idle for a while, since the old install (based on a vendor image) on it stopped booting. Your instructions worked fine and after installation the system sucessfully rebooted to a login prompt. Great :) At that point I got distracted, when I came back a day or so later I found that the Ethernet interface had stopped working. It appeared to be up and had an IP address (meaning it must have passed some packets since the reboot), but no traffic was getting through. Downing and re-upping the interface brought it back to life. I also rebooted the system and found the network interface had come up successfully after the reboot. Then I ran into another problem. I have a SATA hard disk connected by a startech USB to SATA/IDE adapter (I belive its https://www.startech.com/uk/HDD/Adapters/USB-20-to-IDE-or-SATA-Adapter-Cable~USB2SATAIDE though I'm not using the PSU that came with it). When I tried to rsync a large file (an image of the pre-reinstall SD card contents) to said drive it dropped out. I unmounted and re-mounted the drive (which had renamed itself from sda to sdb after the dropout) and tried again, but the same happened again. As usual with this kind of devices, do make sure you have a good power source (for the PINE A64 ideally via the Euler bus instead of the micro-USB connector). Right now i'm using a 5V 4A brick from stontronics. The output from this is split to feed the pine (currently via micro usb) and the HDD. One of the reasons I chose the particular USB to SATA adapter I did is because it is only a data adapter, allowing me to do the power the way I want it. I have just tried a different USB to SATA adapter, this one unfortunately uses a combined data and power connection to the drive, but it has a second USB plug for power, so I chopped up a USB extension and used it to connect said second USB plug direct to the power supply. I figured the pine64 would probably tolerate (and may even benefit from) any backfeeding of it's USB device ports this caused. After the swap I was able to complete the rsync successfully. So it seems that some issue with the USB to SATA adapter I was using (either with the adapter in general or a compatibility issue between the adapter and the pine) was to blame. Connecting external drives without own power or via a powered USB hub is just asking for trouble :) Kind regards, Andrei
Re: Debian Installer on Pine64+ (no serial console)
On 04/05/19 18:49, Andrei POPESCU wrote: On Sb, 04 mai 19, 06:57:43, Andrei POPESCU wrote: I'm ready to start a second try where I will be using manual partitioning (without GPT), hoping I will get a bootable system. This worked, see https://wiki.debian.org/InstallingDebianOn/PINE64/PINEA64 This prompted me to reinstall my pine64 which has been sitting idle for a while, since the old install (based on a vendor image) on it stopped booting. Your instructions worked fine and after installation the system sucessfully rebooted to a login prompt. At that point I got distracted, when I came back a day or so later I found that the Ethernet interface had stopped working. It appeared to be up and had an IP address (meaning it must have passed some packets since the reboot), but no traffic was getting through. Downing and re-upping the interface brought it back to life. I also rebooted the system and found the network interface had come up successfully after the reboot. Then I ran into another problem. I have a SATA hard disk connected by a startech USB to SATA/IDE adapter (I belive its https://www.startech.com/uk/HDD/Adapters/USB-20-to-IDE-or-SATA-Adapter-Cable~USB2SATAIDE though I'm not using the PSU that came with it). When I tried to rsync a large file (an image of the pre-reinstall SD card contents) to said drive it dropped out. I unmounted and re-mounted the drive (which had renamed itself from sda to sdb after the dropout) and tried again, but the same happened again. I have a dmesg log from the second dropout at the bottom of the mail. The running kernel at the time was Linux pine64 4.19.0-4-arm64 #1 SMP Debian 4.19.28-2 (2019-03-15) aarch64 GNU/Linux I then decided to upgrade to linux-image-5.0.0-trunk-arm64_5.0.2-1~exp1, unfortunately that didn't seem to improve things much, rather than the usb SATA dropping out and rsync returning an error, rsync just seemed to hang in uninterruptable sleep with no disk or network IO. [ 987.717823] sd 1:0:0:0: [sdb] tag#0 CDB: Write(10) 2a 00 01 44 e3 c0 00 00 f0 00 [ 987.717830] print_req_error: 256 callbacks suppressed [ 987.717836] print_req_error: I/O error, dev sdb, sector 21291968 [ 988.062499] usb 1-1: USB disconnect, device number 3 [ 988.077085] print_req_error: I/O error, dev sdb, sector 21292208 [ 988.083176] print_req_error: I/O error, dev sdb, sector 21292448 [ 988.089258] print_req_error: I/O error, dev sdb, sector 495192064 [ 988.095381] Buffer I/O error on dev sdb2, logical block 59277312, lost sync page write [ 988.103338] JBD2: Error -5 detected when updating journal superblock for sdb2-8. [ 988.103347] print_req_error: I/O error, dev sdb, sector 21292688 [ 988.110743] Aborting journal on device sdb2-8. [ 988.121233] print_req_error: I/O error, dev sdb, sector 21292928 [ 988.127268] EXT4-fs warning (device sdb2): ext4_end_bio:323: I/O error 10 writing to inode 13 (offset 16777216 size 5246976 starting block 2661632) [ 988.127276] buffer_io_error: 7414 callbacks suppressed [ 988.127285] Buffer I/O error on device sdb2, logical block 39680 [ 988.133319] Buffer I/O error on device sdb2, logical block 39681 [ 988.139341] Buffer I/O error on device sdb2, logical block 39682 [ 988.145368] Buffer I/O error on device sdb2, logical block 39683 [ 988.151390] Buffer I/O error on device sdb2, logical block 39684 [ 988.157416] Buffer I/O error on device sdb2, logical block 39685 [ 988.163438] Buffer I/O error on device sdb2, logical block 39686 [ 988.169467] Buffer I/O error on device sdb2, logical block 39687 [ 988.175489] Buffer I/O error on device sdb2, logical block 39688 [ 988.181516] Buffer I/O error on device sdb2, logical block 39689 [ 988.188059] print_req_error: I/O error, dev sdb, sector 495192064 [ 988.194161] Buffer I/O error on dev sdb2, logical block 59277312, lost sync page write [ 988.202106] JBD2: Error -5 detected when updating journal superblock for sdb2-8. [ 988.202120] print_req_error: I/O error, dev sdb, sector 21293056 [ 988.215542] print_req_error: I/O error, dev sdb, sector 21293296 [ 988.221578] print_req_error: I/O error, dev sdb, sector 21293536 [ 988.228521] EXT4-fs warning (device sdb2): ext4_end_bio:323: I/O error 10 writing to inode 13 (offset 16777216 size 6295552 starting block 2661888) [ 988.230033] EXT4-fs warning (device sdb2): ext4_end_bio:323: I/O error 10 writing to inode 13 (offset 16777216 size 7344128 starting block 2662144) [ 988.231395] EXT4-fs error (device sdb2) in ext4_do_update_inode:5331: Journal has aborted [ 988.239701] EXT4-fs warning (device sdb2): ext4_end_bio:323: I/O error 10 writing to inode 13 (offset 16777216 size 8388608 starting block 2662400) [ 988.240263] Buffer I/O error on dev sdb2, logical block 0, lost sync page write [ 988.247618] EXT4-fs (sdb2): I/O error while writing superblock [ 988.253478] EXT4-fs error (device sdb2): mpage_map_and_submit_extent:2568: comm kworker/u8:1: Failed to mark inode 13 dirty [
openssl sha1 not working with large files on stretch arm.
I got a report from a user about "openssl sha1 " not working for large files on raspbian stretch. I investigated the problem and found that when I tried to use "openssl sha1 " for a large (>2GB) file on Debian armhf stretch or raspbian stretch I get|an error "Value too large for defined data type"| | What puzzles me is that it worked on raspbian buster, and also worked on debian stretch i386. According to the user who reported it to me it worked on Raspbian jessie. Anyone have any clues? I'm not finding anything obvious in the build logs or from a quick look at the source (it seems a couple of source files are built with large file support and the rest are not, but which files doesn't seem to have changed between stretch and buster). |
Re: Rebuilding the entire Debian archive twice on arm64 hardware for fun and proft
On 06/01/19 23:45, Steve McIntyre wrote: In my initial testing for rebuilding armhf only, I did not enable either of these. I was then finding *lots* of "Illegal Instruction" crashes due to CP15 barrier usage in armhf Haskell and Mono programs. This suggests that the baseline architecture in these toolchains is incorrectly set to target ARMv6 rather than ARMv7. That should be fixed and all those packages rebuilt at some point. Haskell does appear to be configured for armv7 in Debian (in the sense that I had to patch the package in Raspbian to make it build for armv6). I would guess for haskell that this is an issue that needs digging into the upstream source, not just looking at the build system.
Re: coturn - bus error on armhf
On 22/12/18 11:33, wf...@niif.hu wrote: Dear ARM porters, I sponsored the 4.5.0.8-1 upload of the coturn package. It's unit test failed with bus error on armhf and sparc64: https://buildd.debian.org/status/package.php?p=coturn The former being a release architecture, I started investigation with that, but I failed to reproduce this on abel.debian.org or harris.debian.org: the package built and the unit test ran fine on both. Could you please advise how to proceed with this problem? The maintainer already requested a giveback, but it hasn't happened yet. Try using an armhf chroot on amdahl.debian.org , iirc armhf chroots on arm64 machines are often stricter about alignment than pure armhf machines.
Re: giveback of knot-resolver 3.1.0-1 on armhf?
On 30/11/18 23:23, Daniel Kahn Gillmor wrote: entry=entry@entry=0xf051c962 Well that looks to be your problem, an unaligned pointer. You need to figure out where it is coming from and fix it.
can't boot 4.18 kernel on wandboard quad.
I recently went to upgrade one of my wandboard quads from 4.9.0-0.bpo.1-armmp to 4.18.0-0.bpo.1-armmp Unfortunately the new kernel doesn't seem to boot, it seems to hang after "starting kernel". Any ideas on what the problem could be? My u-boot environment looks like => mmc dev ${mmcdev} mmc0 is current device => mmc rescan => run loadbootenv 416 bytes read in 28 ms (13.7 KiB/s) => run importbootenv Importing environment from mmc (uEnv.txt)... => printenv baudrate=115200 boot_fdt=try bootcmd=mmc dev ${mmcdev};if mmc rescan; then echo SD/MMC found on device ${mmcdev};if run loadbootenv; then run importbootenv;fi;if test -n $uenvcmd; then echo Running uenvcmd ...;run uenvcmd;fi;if run loadzimage; then run mmcboot;fi;fi; bootdelay=1 bootscript=echo Running bootscript from mmc ...; source console=ttymxc0 ethact=FEC ethaddr=00:1f:7b:b4:05:1a ethprime=FEC fdt_addr=0x1100 fdt_file=imx6q-wandboard.dtb fdt_high=0x filesize=1a0 importbootenv=echo Importing environment from mmc (uEnv.txt)...; env import -t $loadaddr $filesize initrd_addr=0x1300 initrd_high=0x ip_dyn=yes loadaddr=0x1200 loadbootenv=load mmc ${mmcdev}:${mmcpart} ${loadaddr} uEnv.txt loadbootscript=fatload mmc ${mmcdev}:${mmcpart} ${loadaddr} ${script}; loadfdt=load mmc ${mmcdev}:${mmcpart} ${fdt_addr} /dtbs/${fdt_file} loadinitrd=load mmc ${mmcdev}:${mmcpart} ${initrd_addr} initrd.img loadzimage=load mmc ${mmcdev}:${mmcpart} ${loadaddr} zImage mmcargs=setenv bootargs console=${console},${baudrate} ${optargs} root=${mmcroot} rootfstype=${mmcrootfstype} video=${video} mmcboot=run loadfdt; run mmcargs; run loadinitrd; bootz ${loadaddr} ${initrd_addr}:${filesize} ${fdt_addr}; mmcdev=0 mmcpart=1 mmcroot=UUID=b1bbc414-67f6-40c3-a366-d3317a114748 ro mmcrootfstype=ext4 rootwait fixrtc script=boot.scr stderr=serial stdin=serial stdout=serial uimage=uImage => version U-Boot 2013.04-00015-g5cd081a (Jun 23 2013 - 19:58:00) arm-linux-gnueabihf-gcc (crosstool-NG linaro-1.13.1-4.8-2013.06 - Linaro GCC 2013.06) 4.8.2 20130603 (prerelease) GNU ld (crosstool-NG linaro-1.13.1-4.8-2013.06 - Linaro GCC 2013.06) 2.23.2.20130610 Linaro 2013.06 => It's an older (and probablly vendor) u-boot but I would really rather not mess with the u-boot version if I can avoid it as this is a remotely hosted box.
Re: Firefox-esr crashed on ARMHF
On 17/10/18 16:49, ibu ☉ radempa wrote: What I find confusing: The changelogs of firefox-esr for stable and unstable essentially contain the same items (since v52), but the one reproducibly crashes, and the other reproducibly works. Given that I've tried them in the same environment (versions of dependencies, kernel, shell environment), they must be essentially different, or am I wrong? The most obvious suspect in that case would be toolchain issues, particularly the versions of rustc and llvm used to build the packages.
Re: https://lists.debian.org/debian-arm/2018/09/ is no more updated
On 23/09/18 15:20, LinAdmin wrote: The above archive site is broken: "The last update was on 09:00 GMT Fri Sep 21." which is more than 2 days ago while it should regularly updated ... I could be wrong but I think it's only updated if/when new mails arrive.
Is anyone running buster/sid on armel? if so does current udev work?
It has recently been discovered in raspbian ( https://bugs.launchpad.net/raspbian/+bug/1793415 ) that udev doesn't start if built with gcc 8 due to what appears to be a hardening related failure. As a result I have just forced the systemd source package (which builds udev) in raspbian to build with gcc-7. This issue does not appear to affect Debian armhf. Which leaves me wondering about armel, has anyone tried udev 239-8 or 239-9 on armel? if so did it work?
Re: armel/armhf arch qualification for buster: call for DSA, Security, toolchain concernsj
On 18/08/18 11:01, Roger Shimizu wrote: On Sat, Aug 18, 2018 at 12:33 PM, Hideki Yamane wrote: Hi, So it'll be super great if you (or your boss) can kindly sponsor some armel/armhf hardwares that support to install 4GB memory. From DebConf18 seesion "Building a Debian Derivative: Lessons Learned and Solutions Found" by Alex Doyle from Cumulus Network, they run "Jessie on Chromebook" as armel build server https://youtu.be/OPsfX5_YCiQ?t=14m39s Thanks for your info! Maybe we can do same thing for armel, it'd be better than using old NAS with larger memory (I know we're trying to build armel/hf packages on arm64 but just a FYI). Yes. In latest ARM BoF of DebConf18, Steve (93sam) informed us that he's trying to rebuild all the armhf archive on arm64 box. When that's done, we will get to know how many packages need to fix. If the number of those packages is limited and can be fixed before buster, we can choose this way for both armhf and armel. But if the number is huge that we don't have enough manpower to fix, we need to find real machine (such as rack-mounted armhf/armel NAS box) to work as buildd. Personal opinion but (as a derivative maintainer) I really think DSA are being unreasonable here. A buildd does not need to be some critical 100% uptime server, if it occasionally goes down to the point where it can't be recovered remotely then so be it. Arm ports aside I don't think having "server class" hardware available should be a blocker for getting or maintaining a port in Debian. It never has been in the past and I don't see what has changed now. And I think with a serial console and remote power most issues can be resolved remotely. We have been running wandboard quads for raspbian for years and I have never hard a situation I couldn't get out of with remote power and serial console. On the subject of building on arm64 hardware we now do this for some of our builds in raspbian. It *mostly* works but we do see some testsuite failures that we don't see on the wandboards (otoh we have some testsuites that pass on the arm64 box but not the wandboards, go figure) and I have also been having a build hang with rust recently, but other than that I haven't noticed any prblems.
Re: causes for this?
On 24/06/18 14:44, Alan Corey wrote: Yeah, texlive is like 1 GB. I used it once about 10 years ago but I have it on every Debian or Raspbian machine. I wish there was more flexibility in dependencies somehow. The large size is caused by a couple of factors. 1. Debian bundles together tex packages into a small number of Debian packages, I understand why they do this (tex packages are mostly quite small and there is an overhead for each package in the Debian archive) but it basically means that you normally end up installing all the packages because you end up wanting something from each. 2. Up to and including stretch texlive-latex-base, texlive-latex-recommended and texlive-latex extra declared "recommends on" their corresponding doc packages. These doc packages (especially texlive-latex-extra-doc) are unfortunately very large adding up to about 450 megabytes of download and about 550 megabytes of installed size). You can avoid the second issue by telling apt not to install recommends, unfortunately I don't think there is much you can do about the first.
Re: Anybody here know what a rock64 is?
On 20/04/18 13:33, Alan Corey wrote: Well yes, all of that's true, but would raspberrypi.org mind if Debian borrowed a copy of their scripts? It's MIT licensed so it should be fine to borrow https://github.com/RPi-Distro/raspi-config/LICENSE raspi-config isn't 3b specific Indeed, it's designed for the whole Pi family. Some parts of it are generic other parts are specific to either the Raspberry Pi hardware/firmware or to the way raspberry pi have set up their image. The locales part is generic, but all it's doing is calling "dpkg-reconfigure locales" so there isn't much to borrow there.
gcc-8, libatomic and armel/raspbian
It seems that gcc-8 unconditionally builds parts of libatomic with "-march=armv7-a+fp". This was detected by Raspbian's armv7 contamination checker before the package was uploaded to raspbian. There is no such checking for Debian armel so the package was accepted there and is now being shipped in Debian armel sid and buster. I patched out the flag in the raspbian package and things still seem to build successfully with the resulting package passing our contamination checker. A debdiff should appear soon at http://debdiffs.raspbian.org/main/g/gcc-8/ It's possible (even likely) that this is a false positive and that the code in question is behind runtime detection, even if it is a false positive though problematic to have armv7 tags on a static library because they have the potential to cause further false positives in downstream code. Anyone have any thoughts?
Re: all translations for Jessie are 403.
On 09/11/17 06:22, Gene Heskett wrote: On Thursday 09 November 2017 00:54:52 Paul Wise wrote: On Fri, Nov 3, 2017 at 8:54 PM, Gene Heskett wrote: I did an update on an armhf/jessie install last night, and noticed that all the files for translations were missing, which resulted in quite a bit of verbosity from synaptic. Its been this way for in excess of a week. You will need to be more specific; which sources.list are you using, what errors are you getting etc. This sources.list: pi@picnc:~ $ cat /etc/apt/sources.list deb http://mirrordirector.raspbian.org/raspbian/ jessie main contrib non-free rpi # Uncomment line below then 'apt-get update' to enable 'apt-get source' #deb-src http://archive.raspbian.org/raspbian/ jessie main contrib non-free rpi And in /etc/apt/sources.list.d pi@picnc:/etc/apt/sources.list.d $ cat raspi.list deb http://archive.raspberrypi.org/debian/ jessie main ui # Uncomment line below then 'apt-get update' to enable 'apt-get source' #deb-src http://archive.raspberrypi.org/debian/ jessie main ui And pi@picnc:/etc/apt/sources.list.d $ cat linuxcnc-buildbot.list #debhttp://buildbot.linuxcnc.org/ jessie master-sim #deb-srchttp://buildbot.linuxcnc.org/ jessie master-sim deb http://buildbot.linuxcnc.org/ jessie master-sim Just now its upgrading about 8 or 9 files but isn't fussing about translations. Its not doing the kernel because its a special kernel I have pinned. And as I said, apparently all the translations are missing. And tonight, because you asked for the errors, it all worked on the rpi3. And thats the only jessie install. Go figure. But then I was running apt over an ssh login. Not synaptic because some "genius" has set up a security rule someplace I haven't yet found and nuked that keeps me from running synaptic-pkexec over an ssh -Y session. What puzzles me is I don't think raspbian has ever shipped translations, so I wonder what has changed recently.
Re: how to distinguish armel and armhf at runtime?
On 23/09/17 19:42, Uwe Kleine-König wrote: On Sat, Sep 23, 2017 at 02:08:28AM +0100, Steve McIntyre wrote: On Fri, Sep 22, 2017 at 09:57:02PM +0200, Uwe Kleine-König wrote: Hello, for the package sparse I have to distinguish armel and armhf. The reason is that sparse parses system headers and so has to know which cpp symbols to define. Usually it uses uname -m to distinguish platforms but that isn't suitable to tell armel and armhf apart. For armhf I need to define __ARM_PCS_VFP but that must be absent on armel. Right, that makes sense. For upstreaming a fix it would be great if the test would not be Debian specific. That's likely to be difficult - it's perfectly possible to have headers etc. for both on a single system with multi-arch (for example). It's an ABI choice, not a hardware difference. What are you trying to work this out for? Package build time? No, it's for run time. The target is sparse which can parse C and to be able to handle system includes it needs the right cpp variables defined. Are people expected to have gcc installed when using this tool? if so gcc -dumpmachine may be an option.
Bug#850906: ceph FTBFS on armel, libtool/linker related issues.
Package: ceph Version: 10.2.5-5 Severity: serious x-debbugs-cc: debian-arm@lists.debian.org The most recent upload of ceph fixed the build on most architectures but unfortunately armel is still failing. libtool: relink: g++ -fPIC -DPIC -shared -nostdlib /usr/lib/gcc/arm-linux-gnueabi/6/../../../arm-linux-gnueabi/crti.o /usr/lib/gcc/arm-linux-gnueabi/6/crtbeginS.o java/native/.libs/libcephfs_jni_la-libcephfs_jni.o java/native/.libs/libcephfs_jni_la-JniConstants.o -Wl,--whole-archive ./.libs/libcommon.a -Wl,--no-whole-archive -Wl,--as-needed -L/«PKGBUILDDIR»/debian/tmp/usr/lib/arm-linux-gnueabi -L/usr/lib/arm-linux-gnueabi -lcephfs -ldl -lboost_thread -lboost_random -lblkid -luuid -lrt -lboost_iostreams -lboost_system -L/usr/lib/gcc/arm-linux-gnueabi/6 -L/usr/lib/gcc/arm-linux-gnueabi/6/../../../arm-linux-gnueabi -L/usr/lib/gcc/arm-linux-gnueabi/6/../../.. -L/lib/arm-linux-gnueabi -lstdc++ -lm -lc -lgcc_s /usr/lib/gcc/arm-linux-gnueabi/6/crtendS.o /usr/lib/gcc/arm-linux-gnueabi/6/../../../arm-linux-gnueabi/crtn.o -O2 -g -fstack-protector-strong -g -O2 -fstack-protector-strong -Wl,-z -Wl,relro -Wl,-z -Wl,now -Wl,-z -Wl,relro -Wl,-z -Wl,now -Wl,-soname -Wl,libcephfs_jni.so.1 -o .libs/libcephfs_jni.so.1.0.0 /usr/bin/ld: /«PKGBUILDDIR»/debian/tmp/usr/lib/arm-linux-gnueabi/libcephfs.so: invalid string offset 50471 >= 3194 for section `.strtab' <--snip--> libtool: warning: remember to run 'libtool --finish /usr/lib/arm-linux-gnueabi' make[5]: Leaving directory '/«PKGBUILDDIR»/src' Makefile:32173: recipe for target 'install-am' failed make[4]: *** [install-am] Error 2 make[4]: Leaving directory '/«PKGBUILDDIR»/src' Makefile:30865: recipe for target 'install-recursive' failed make[3]: *** [install-recursive] Error 1 make[3]: Leaving directory '/«PKGBUILDDIR»/src' Makefile:32167: recipe for target 'install' failed make[2]: *** [install] Error 2 make[2]: Leaving directory '/«PKGBUILDDIR»/src' Makefile:703: recipe for target 'install-recursive' failed make[1]: *** [install-recursive] Error 1 make[1]: Leaving directory '/«PKGBUILDDIR»' dh_auto_install: make -j4 install DESTDIR=/«PKGBUILDDIR»/debian/tmp AM_UPDATE_INFO_DIR=no returned exit code 2 debian/rules:73: recipe for target 'binary-arch' failed This seems like a rather strange error, putting debian-arm in cc in case anyone there has any ideas.
Re: Help need with build failure of ceph 10.2.5-2 on armel
On 26/12/16 21:52, Gaudenz Steinlin wrote: AFAICS the build failure appears if a source file uses "#include". AFAIU the build tries to build a plugin with NEON instructions to be used depending on runtime detection. You need to add "-mfloat-abi=softfp" to the compilation of the neon-using plugin on armel. This flag must only be used on armel (not on armhf) and must only be used for code that is gaurded behind appropriate runtime detection.
Re: cubox i4x4 hard drive seen in D-I but not in installed system.
On 20/12/16 12:50, Gunnar Wolf wrote: I have a CuBox-i 4 (not 4x4). It's currently in stable use, and have not fiddled with it for long (and right now, it sits 7000Km away from me), but if I recall correctly, any USB media needed for the boot process (i.e. a keyboard) has to be inserted in the lower port. I know the issue seems completely different from yours, but this might be an important point to get your issue properly solved. Just tried moving the power cable for my HDD to the lower port, that made no difference. I then tried plugging the power cable for hard drive into a HTC phone charger. That made the drive show up without needing to use the "hotplug" option. So it seems this is indeed a timing issue between power up of the USB port and initialisation of the SATA interface either because USB init is being sequenced after SATA init or because the hard drive takes some time from power up to being ready to talk on the SATA interface.
Re: cubox i4x4 hard drive seen in D-I but not in installed system.
On 13/12/16 21:42, peter green wrote: I would guess at this point either a race condition or a power glitch (maybe powering the HDD off one of the USB ports wasn't such a good idea). OK, I found that adding ahci-imx.hotplug=1 to the kernel command line made it work. Checking the schematic I notice that the USB ports seem to have power control, I wonder if they are power-cycled at some point during the boot process.
Re: cubox i4x4 hard drive seen in D-I but not in installed system.
On 13/12/16 19:59, peter green wrote: I have just bought a cubox i4x4 and installed Debian Jessie on it using D-I (concatenatable netboot). I am using a SD card for the rootfs and plan to use a hard drive to store chroots. Unfortunately while I can see the hard drive in D-I I can't see it after installing. The ahci_imx module seems to be loaded but sda doesn't appear in /dev. I upgraded to the kernel from backports. This made no difference. Any idea how to go about troubleshooting this issue? the box isn't much use to me if I can't make the hard drive work. Ok, after some discussions on irc I discovered some more. It seems that the system sees the port but fails to detect the drive. [2.694758] ahci-imx 220.sata: fsl,transmit-level-mV value 1104, using 0044 [2.694778] ahci-imx 220.sata: fsl,transmit-boost-mdB value 0, using [2.694791] ahci-imx 220.sata: fsl,transmit-atten-16ths value 9, using 2000 [2.694802] ahci-imx 220.sata: fsl,receive-eq-mdB not specified, using 0500 [2.698172] ahci-imx 220.sata: SSS flag set, parallel bus scan disabled [2.698208] ahci-imx 220.sata: AHCI 0001.0300 32 slots 1 ports 3 Gbps 0x1 impl platform mode [2.698223] ahci-imx 220.sata: flags: ncq sntf stag pm led clo only pmp pio slum part ccc apst [2.699973] scsi host0: ahci-imx [3.014679] ahci-imx 220.sata: no device found, disabling link. [3.014690] ahci-imx 220.sata: pass .hotplug=1 to enable hotplug However if I unload and reload the driver then it sees the drive successfully. I would guess at this point either a race condition or a power glitch (maybe powering the HDD off one of the USB ports wasn't such a good idea).
cubox i4x4 hard drive seen in D-I but not in installed system.
I have just bought a cubox i4x4 and installed Debian Jessie on it using D-I (concatenatable netboot). I am using a SD card for the rootfs and plan to use a hard drive to store chroots. Unfortunately while I can see the hard drive in D-I I can't see it after installing. The ahci_imx module seems to be loaded but sda doesn't appear in /dev. I upgraded to the kernel from backports. This made no difference. Any idea how to go about troubleshooting this issue? the box isn't much use to me if I can't make the hard drive work. When trying to debug I wondered if the wrong dtb was being used. Unfortunately I can't seem to figure out what puts the dtb in /boot and how it selects which dtb to use.
Re: Any chance to provide upstream access to armhf machine thats comparable to our autobuilder?
On 13/12/16 07:37, Andreas Tille wrote: Hi, when discussing bug #800469 upstream thinks that armhf should work and the question came up whether there is some chance to access an armhf machine that is comparable to our autobuilders. The porterbox abel.debian.org runs the same hardware as the buildds and has both armel and armhf chroots available. There is a procedure for guest accounts on porterboxes, though I haven't used it myself so I don't know how well it works in practice.
Re: What current NAS platform is recommended for installing Debian?
On 03/12/16 15:53, Victor Hooi wrote: Hi, I'm looking at buying a cheap 2-bay NAS to install Debian onto. The QNAP TS-21X units are quite good, but are a little dated. armel's days as a full Debian port appear to be coming to an end (it looks like stretch will be the last release). There is talk of partial architectures but partial architectures have been just talk for years. I then saw the Seagate Personal Cloud - however, apparently these are also a bit dated: A bit dated perhaps but nowhere near as dated as the QNAPs.
Re: postgis failing to build on arm64 only
On 12/10/16 18:57, Wookey wrote: That seems quite likely. It might just be the senstivity of C++ symbols to compiler changes. Unfortunately so far as I know one can't test such a thing on the porter boxes (because there is no way to unstall the local gdal you just built due to lack of root rights). Yeah, I ran into the same thing when trying to sort out freepascal on powerpc. In the end I resorted into using qemu (which worked but is bloody slow). I wonder if it could be worth talking to DSA about using container/namespace technology to improve the porterbox experiance.
Re: arm64 pointer tagging, VA-bits, the end of the world...
On 17/08/16 17:06, Leif Lindholm wrote: Indeed. And we need to stay on the ball with that, and try to ensure any changes we do from this point onwards are at least 56-bit safe. And start agitating against pointer tagging in general. AIUI a big problem is that some systems (in particular javascript engines), don't merely want to tag pointers, they want to create a 64-bit "compact variant" that stores multiple different types of data including double precision integers and pointers (possiblly even more than one type of pointer) using non-overlapping bit patterns. This is possible due to the fact that the double precision floating point format reserves a large chunk of code space (2^53 bit patterns) for NaNs and infinities. Assuming your application doesn't need a crazy number of different NaNs these bit patterns can be repurposed for other things. If your pointers are 48 bit then it's no problem to store multiple different types of pointer and other values in the NaN-space. On the other hand if your pointers are 56 bit then you have a serious problem. Alignment considerations mean that if the compact variant had to grow beyond 64 bit it would probablly end up being 128 bit which is a fairly steep penalty on memory usage.
Re: Which port for armv4l?
On 17/08/16 10:19, Adam Wysocki wrote: Is it possible for a compiler to generate BX instructions without Thumb code (for example to return from a function with bx lr, where lr will always be even - no Thumb)? Correct me if I'm wrong, but I thought that BX patch (to emulate BX instruction in kernel) was supposed to allow running code with BX instructions on a processor that lacks it, and that's why I want to use it (to run Debian for armel on StrongARM SA-1100, which has ARMv4 core, so without Thumb). Code in Debian for armel architecture doesn't use Thumb? Unless explicitly told otherwise the compilers in Debian armel build arm mode code. I don't think there is anything forbidding packagers from chosing thumb mode for their code but they would have to go out of their way to make that choice and i'm not aware of anyone doing so. However the compiler doesn't know whether you will be mixing the arm code it built with thumb code built with another compiler invocation. So it uses BX for function calls and returns.
Re: Broadcom BCM2709, ARMv8, and missing CPU features
On 28/07/16 17:35, Gunnar Wolf wrote: I'm far from an absolute expert in this area... But I am fairly certain of what I say — That is, I have a RPi 1 and 2B, and they cannot boot from the same images. That depends what is in the image. The current raspberry pi firmware works on all pi models (older firmware will only work with older pi models). The Pi1 needs a specific kernel. The Pi2 and the Pi3 can run the same 32-bit kernel (at least with foundation kernels, I dunno what the situation is with upstream kernels) The firmware by default selects a suitable kernel (kernel.img or kernel7.img) and device tree (each Pi model has a different one though IIRC they are pretty similar) based on the detected hardware. There is some experimental 64-bit kernel/bootloader stuff out there for the raspberry pi 3 https://www.raspberrypi.org/forums/viewtopic.php?f=72=137963=aarch64 https://www.raspberrypi.org/forums/viewtopic.php?f=72=143765 Userland obviously has to be compatible with the hardware and kernel. Raspbian or Debian armel userlands should be usable on any Pi model. Debian armhf userland should be usable on a pi2 or pi3 but clearly not a pi1. Debian arm64 will obviously require a 64-bit kernel and a pi3.
Re: Summary of the ARM ports BoF at DC16
On 22/07/16 02:36, Steve McIntyre wrote: Affordable, usable machines are available now, e.g. the Cello Does anyone know what is going on with this. Have boards actually started shipping?
SD card device name changed on wandboard quad.
While upgrading a wandboard quad from 4.4 to 4.6 (debian armmp kernels) I got a boot failure. I tracked this down to the device name for the SD card changing from mmcblk0 to mmcblk2 Any idea why this has changed? There don't seem to be any other lower-numbered mmcblk devices.
Re: armhf ABI flag problems with fpc-built binaries
On 09/06/16 18:55, Steve McIntyre wrote: Hi folks, I'm one of the ARM porters, and I've recently run a scan of binaries in the archive to check on the state of the binaries for armel and armhf. As part of the ARM ABI, binaries (libraries and programs) are expected to specify ELF flags to specify whether they're using the hard-float or soft-float ABI (if they care specifically - some don't). I've found a few fpc-using packages (both under your maintenance and not) that appear to get this wrong, which suggests that there might be a toolchain issue here in fpc and friends. My scanner is telling me that the following armhf packages are broken: doublecmd-plugins_0.7.1-2_armhf.debhf_flags_wrong:6 fp-compiler-3.0.0_3.0.0+dfsg-4_armhf.deb hf_flags_wrong:4 fp-ide-3.0.0_3.0.0+dfsg-4_armhf.debhf_flags_wrong:1 fp-utils-3.0.0_3.0.0+dfsg-4_armhf.deb hf_flags_wrong:28 no_hf_flags:2 gearhead_1.300-1_armhf.deb hf_flags_wrong:1 pasdoc_0.14.0-1_armhf.deb hf_flags_wrong:1 "hf_flags_wrong" here means that the package is targeting armhf, but binaries within it claim to use the soft-float ABI. "no_hf_flags" is the number of binaries that have no ABI float flags attached. To see for yourself, you can use "readelf --file-header" to inspect the flags on binaries - look for Flags: 0x5000202, has entry point, Version5 EABI, soft-float ABI etc. armhf binaries should show Flags: 0x5000402, has entry point, Version5 EABI, hard-float ABI instead. I'm not sure exactly where fpc would be setting flags like this (if anywhere) or if it's calling binutils incorrectly maybe. Can anybody help me dig into this please? (See the attached list for details of the broken files.) Most likely the place to start in the fpc code is compiler/arm/agarmgas.pas If someone can tell me what needs to be done at an assembler level I can look into making fpc do it. (unfortunately infinity's paste referenced in the bug report has expired).
Re: kexec-tools installs on arm64
On 13/06/16 14:49, Phil Endecott wrote: Dear All, I was surprised to find that "apt-get install kexec-tools" succeeds on an arm64 system - it installs the armhf version. It then fails to run, with an "Unsupported machine type" error. Is this an error in the multiarch tagging for the kexec-tools package? I would class it as a design limitation of multiarch. apt has no way of knowing if a particular foreign-architecture package will actually be useful. Only what dependencies it will satisfy and whether versions for multiple architectures can be co-installed.
Re: [Stretch] Status for architecture qualification
On 07/06/16 19:38, Martin Michlmayr wrote: * Steve McIntyre[2016-06-06 15:14]: However, I will admit (again) that armel is starting to lose upstream support in some cases. I'm tempted to suggest that Stretch should be the last release for armel for that reason. Which upstream problems do you see? A big concern going forward is C++11 atomics. AIUI (unless something has changed recently) these are unimplemented on armel causing code that uses them to FTBFS. Right now very little code uses them but if a major peice of infrastructure starts using them it could put armel in a very sticky sitution. armv4t doesn't really have good atomic instructions (there is "swp" but it has to be emulated on later architectures and i'm not sure it's enough to support the normal range of operations) . AIUI there are kernel helpers available but it would need someone with good knowlage of the subject to implement the C++11 atomics in terms of the kernel helpers.
Re: [Stretch] Status for architecture qualification
On 05/06/16 23:01, Wookey wrote: Ok, so a total of 6 shared between the two architectures? Thats what it looks like to me. And I think we are building armhf on the arm64 build machines too? I don't think so. At least I see no evidence of it on buildd.debian.org .
Re: [Stretch] Status for architecture qualification
On 05/06/16 11:01, Niels Thykier wrote: all arm ports have DSA concerns. Is there a current reference to what these concerns are? Is there still a lack of out of band management? (the old mail I found on the topic said it was "being worked on", sledge whats the status here?) are there other concerns? - armel has a RT concern about lack of buildds (only 2) I think this is outdated, my understanding is that armel and armhf are now using a shared buildd pool. I see arnold, hoiby, henze, hasse, antheil and hartmann recently active on the armel buildd page and the same on the armhf page.
Re: [Stretch] Status for architecture qualification
On 05/06/16 13:00, Holger Levsen wrote: On Sun, Jun 05, 2016 at 01:26:39PM +0200, John Paul Adrian Glaubitz wrote: ppc64: This architecture is basically on par with the release architectures. We have over 11.000 packages installed [...] sparc64: We are close to 11.000 installed packages. I'm not sure whether you are talking about source or binary packages but sid/amd64 has over 24000 source packages and over 5 binary packages, so I would call the above "on par". Or what am I missing? I presume he is talking about source packages in the "installed" state in wana-build. For comparison this is currently 12153 for amd64 and 10937 for mips. (this leads to the interesting conclusion that about half the source packages in sid are arch-all only)
Re: affordable arm boards with plenty of ram.
On 19/05/16 18:35, Vagrant Cascadian wrote: Firefly-4GB: price: ~US$260 kernel: 4.6.x in experimental (4.4.x worked fine, but 4.5.x broke ethernet). cpu: quad-core cortex-a12?/a17? (Rockchip rk3288) ram: 4GB disk: USB2 u-boot: patched u-boot 2016.01 to recognize 4GB of ram builds: 182 This looks interesting, do you have more information on it?
Re: affordable arm boards with plenty of ram.
On 19/05/16 17:25, Phil Endecott wrote: peter green<plugwash p10link.net> writes: As much ram as possible. Plenty of CPU power would be nice. A good storage interface (in order of preference SATA is better than USB3 which is better than USB2). Support in Debian kernels would be nice. arm64 support would be nice. This is what we've all been waiting for for years. Indeed. I figured I probablly wasn't the only person on this list waiting for such a thing and hence I put out this message asking if anyone had spotted something I hadn't. I also asked on irc. Currently looking at the cubox i4x4. Anyone have any other suggestions? No, they're all deficient in some way. That one (a) has a Freescale processor, which I would avoid after my experience with the i.MX53, mmm, I did have a nitrogen6x die on me (out of warranty and given where I live sending it back to boundary for an opinion on why it failed and whether it was repairable would have been prohibitively expensive). Other than that i've been pretty happy with imx6 based stuff. (b) isn't 64-bit, (c) has only 1 GHz CPU clock. You can get faster CPUs (4x 2GHz A15, so probably about 3X the speed) but half the RAM and no SATA (but USB3) in the ODROID-XU4. Yeah, that is pretty much my conclusion. For the ammount of money I'm prepared to spend on arm boards you can get modern CPU cores of 4GB of ram but not both. Theres the nvidia jetson TX1 but it costs far more than I personally consider reasonable (though I would note that sili and i'm not sure what the kernel situation with it is (it apparently has some support in mainline but I dunno if it has enough to be usable). Theres the husky/cello (which is overbudget but good enough specs that I would consider blowing said budget) but afaict it's not actually available yet. If you want 64-bit, the ODROID-C2 which I have is possibly the best option - but it has A53 not A57 cores (so maybe 2X the speed of the Cubox 4x4), and 2 GB RAM, and no SATA. The question of whether CPU, RAM or filesystem I/O limits performance in practice is an interesting one. The feeling I have is the for kernel compiles my ODROID-C2 is CPU-limited. The thing with ram is that most of the time it doesn't matter much but when it does matter it really matters. Grinding swap sucks. Let us know what you decide. If noone suggests anything better by the end of today I think i'll go ahead and order the cubox i4x4
Re: [llvm-dev] llvm-toolchain-3.8 on lower arm targets, specifically Debian armel and Raspbian.
On 18/05/16 04:50, Tim Northover wrote: If you don't need/want the various Sanitizer runtimes (e.g. you don't support sanitizers or already have versions provided with GCC) then it's as easy as not downloading compiler-rt or removing it from the projects/ directory before running CMake. The build should carry on quite happily without it. If you do need the sanitizers from compiler-rt, you can disable the bits that are causing problems right now with "cmake -DCOMPILER_RT_BUILD_BUILTINS=OFF". Of course, the sanitizers themselves may suffer from similar issues -- you just won't know until you get there. Thanks, -DCOMPILER_RT_BUILD_BUILTINS=OFF did the trick and I got a successful build for raspbian that passed my armv7 contamination checker script (which is not 100% foolproof but it's the best we have) and uploaded it to raspbian stretch.
affordable arm boards with plenty of ram.
I'm looking for an arm board/box to add to my collection of buildboxes. Critera (in roughly descending order of importance): Reasonablly affordable (upper limit ~£200) As much ram as possible. Plenty of CPU power would be nice. A good storage interface (in order of preference SATA is better than USB3 which is better than USB2). Support in Debian kernels would be nice. arm64 support would be nice. Currently looking at the cubox i4x4. Anyone have any other suggestions?
Re: Hybrid armhf/arm64 system
On 19/05/16 01:06, Steve McIntyre wrote: Yes, there is - #822489 in ldconfig. It's fixed already and that fix should be migrating into testing any day. According to the pts it's blocked by a "block-udeb"
Re: [llvm-dev] llvm-toolchain-3.8 on lower arm targets, specifically Debian armel and Raspbian.
On 17/05/16 22:38, Tim Northover wrote: Compiler-rt is the equivalent of libgcc, and Clang can use the existing host's libgcc quite happily so it's really not that important unless you're trying to build a GNU-free environment for whatever reason. Thanks Can you tell me how I would go about telling the cmake buildsystem (which is what the Debian packaging currently seems to use) that I want it to build without compiler-rt?
Re: [llvm-dev] llvm-toolchain-3.8 on lower arm targets, specifically Debian armel and Raspbian.
On 17/05/16 18:07, Tim Northover wrote: Yes, it looks like we'd need to conditionally compile these functions in ARM mode and use the v6 barrier instead of dmb ("mcr p15, #0, r0, c7, c10, #5" I believe) to support the ARM1176JZF-S in RPi. You'd probably also want the build system to use an explicit -march=armv6 or something so you're not at the mercy of how the host compiler was configured. I suspect you're the first person to try and get compiler-rt going for pre-v7 hardware, which most of us prefer to pretend never happened most of the time. Thank you very much for your reply. Can you clarify what exactly compiler-rt is and how important it is to the stack as a whole. It seems that llvm-toolchain-3.7 built successfully on raspbian without building compiler-rt while llvm-toolchain-3.8 built an armv7 compiler-rt (and hence got blocked from entering the raspbian repos).
llvm-toolchain-3.8 on lower arm targets, specifically Debian armel and Raspbian.
llvm-toolchain-3.8 seems to have problems on debian armel and raspbian. On raspbian it builds but our armv7 contamination checker blocked it from entering the repo. Further investigation showed that "compiler-rt" was being built with -march=armv7 . I was able to remove the -march with some build-system hacker but then I got a failure on projects/compiler-rt/lib/builtins/arm/sync_fetch_and_add_4.S [ 8%] Building ASM object projects/compiler-rt/lib/builtins/CMakeFiles/clang_rt.builtins-armhf.dir/arm/sync_fetch_and_add_4.S.o cd /llvm-toolchain-3.8-3.8/build-llvm/projects/compiler-rt/lib/builtins && /usr/bin/gcc-5 -D_GNU_SOURCE -D__STDC_CONSTANT_MACROS -D__STDC_FORMAT_MACROS -D__STDC_LIMIT_MACROS -I/llvm-toolchain-3.8-3.8/build-llvm/projects/compiler-rt/lib/builtins -I/llvm-toolchain-3.8-3.8/projects/compiler-rt/lib/builtins -I/llvm-toolchain-3.8-3.8/build-llvm/include -I/llvm-toolchain-3.8-3.8/include -fPIC -O2 -g -DNDEBUG -mfloat-abi=hard -o CMakeFiles/clang_rt.builtins-armhf.dir/arm/sync_fetch_and_add_4.S.o -c /llvm-toolchain-3.8-3.8/projects/compiler-rt/lib/builtins/arm/sync_fetch_and_add_4.S /llvm-toolchain-3.8-3.8/projects/compiler-rt/lib/builtins/arm/sync_fetch_and_add_4.S: Assembler messages: /llvm-toolchain-3.8-3.8/projects/compiler-rt/lib/builtins/arm/sync_fetch_and_add_4.S:20: Error: selected processor does not support `dmb' in Thumb mode /llvm-toolchain-3.8-3.8/projects/compiler-rt/lib/builtins/arm/sync_fetch_and_add_4.S:20: Error: selected processor does not support `ldrex r0,[r12]' in Thumb mode /llvm-toolchain-3.8-3.8/projects/compiler-rt/lib/builtins/arm/sync_fetch_and_add_4.S:20: Error: cannot honor width suffix -- `add r2,r0,r1' /llvm-toolchain-3.8-3.8/projects/compiler-rt/lib/builtins/arm/sync_fetch_and_add_4.S:20: Error: selected processor does not support `strex r3,r2,[r12]' in Thumb mode /llvm-toolchain-3.8-3.8/projects/compiler-rt/lib/builtins/arm/sync_fetch_and_add_4.S:20: Error: selected processor does not support `dmb' in Thumb mode projects/compiler-rt/lib/builtins/CMakeFiles/clang_rt.builtins-armhf.dir/build.make:1097: recipe for target 'projects/compiler-rt/lib/builtins/CMakeFiles/clang_rt.builtins-armhf.dir/arm/sync_fetch_and_add_4.S.o' failed make[4]: *** [projects/compiler-rt/lib/builtins/CMakeFiles/clang_rt.builtins-armhf.dir/arm/sync_fetch_and_add_4.S.o] Error 1 make[4]: Leaving directory '/llvm-toolchain-3.8-3.8/build-llvm' CMakeFiles/Makefile2:12146: recipe for target 'projects/compiler-rt/lib/builtins/CMakeFiles/clang_rt.builtins-armhf.dir/all' failed make[3]: *** [projects/compiler-rt/lib/builtins/CMakeFiles/clang_rt.builtins-armhf.dir/all] Error 2 make[3]: *** Waiting for unfinished jobs... I traced this back to code in compiler-rt/lib/builtins/arm/sync-ops.h . The comments talk about both arm mode and thumb2 mode code but I only actually see thumb2 code in the file. I'm kinda stuck here. I expect this would be a problem for armel too (though it may go unnoticed due to the fact the armel autobuilders now run on armv7 hardware and afaict Debian armel unlike raspbian doesn't have any contamination checking in place) but the build doesn't even get that far. It fails earlier with. /«PKGBUILDDIR»/include/llvm/Support/ThreadPool.h: In member function 'std::shared_future llvm::ThreadPool::async(Function&&, Args&& ...)': /«PKGBUILDDIR»/include/llvm/Support/ThreadPool.h:77:77: error: return type 'class std::shared_future' is incomplete inline std::shared_future async(Function&, Args&&... ArgList) { ^ /«PKGBUILDDIR»/include/llvm/Support/ThreadPool.h: In member function 'std::shared_future llvm::ThreadPool::async(Function&&)': /«PKGBUILDDIR»/include/llvm/Support/ThreadPool.h:95:57: error: return type 'class std::shared_future' is incomplete inline std::shared_future async(Function&) {
Re: Hybrid armhf/arm64 system
On 14/05/16 23:13, Alan Corey wrote: On 5/14/16, Wookeywrote: I consider the stable/testing choice/tradeoffs to be exactly the same for arm and x86. So it depends what you are using the box for. I think it's worth bearing in mind the size of the userbase here, something in arm probably doesn't get tested as much as something in i386. Same amount of time maybe but less actual usage. That is true, and hence the chance of being the first person to hit a regression is going to be higher on arm* testing than on i386/amd64 testing. On the other hand while arm64 scraped into the jessie release it only barely did so. At the time of the jessie release the *only* arm64 hardware available to us was internal arm devboards and a bit of server hardware. So afaict arm64 jessie never really got any end-user testing.
Re: ODROID-C2
On 06/05/16 20:59, Phil Endecott wrote: It interests me to speculate as to why, at both a technical and an organisational level, these people are doing their own things, rather than making "real" Debian work on this board. Are any of the people involved in this even reading this list? To get "plain Debian" working on a board with a new SoC requires a LOT of effort upstreaming patches (since the Debian maintainers understandablly don't want to put random vendor crap in their packages and try to support it locally). Even when it does work there are often functional limitations due to code that is either non-free or is free but needs major reworking before upstream will take it. Whereas to get a mostly Debian system with a few custom bits (kernel, bootloader, possiblly custom versions of a few libraries) going is much easier.
Re: Supporting armel/armhf in wheezy-lts
On 23/04/16 13:41, Raphael Hertzog wrote: Hi, On Fri, 22 Apr 2016, Tollef Fog Heen wrote: I am not speaking on behalf of DSA here. Thanks for making this clear. I also want to explain why I included DSA in the discussion: I wanted to make sure that the fact that we run wheezy armel/armhf buildd for two more years do not go against some DSA plans to decommission some machines running those buildd and that you had no other problems to keep those machine running during the LTS timeframe. I'm not DSA but my understanding is that all suites are built on the same buildds, since jessie supports armel and armhf and squeeze will almost certainly support at least one and hopefully both of those we will need autobuilder hardware for the forseeable future.
Re: [Pkg-julia-devel] Bug#820220: Bug#820220: Bug#820220: Bug#820220: julia: FTBFS on armel/armhf
On 08/04/16 20:55, Graham Inggs wrote: Secondly, it should be possible to reproduce the fault by building in an armhf schroot on abel.d.o, which has the same hardware as the buildds. A temporary solution would be to upload armhf binaries built on asachi.d.o. It would be useful if someone can reproduce the issue and get a dissasembly of the failure location. My gut feeling is that llvm is trying to use neon without first checking it is supported. Afaict the marvell boards we are using for armhf buildds (and for the abel.d.o porterbox) don't have neon. [1] https://buildd.debian.org/status/package.php?p=llvm-toolchain-3.8=unstable
Re: [Reproducible-builds] Raspi 3 suitable for arm64?
On 06/03/16 16:34, Vagrant Cascadian wrote: I got the impression it was another broadcom chip, maybe with a cortex-a53? Gotta love ARM's confusing namespace. It is indeed a broadcom chip. It's like the Pi2 but with a quad cortex A53 complex instead of the quad cortex A7 complex and with a substantially bumped up clockspeed. Ram is still limited to 1GB by limitations in the videocore bus strutucture. Right now only a 32-bit kernel is available from raspberry pi. There is a config.txt option to make the cores start in 64-bit mode ( https://www.raspberrypi.org/forums/viewtopic.php?f=72=137963#p915771 ) but it's not clear if Raspberry Pi will put in the effort to do a 64-bit port of the platform code in the kernel and if not whether someone else will.
Re: Bug#790925: pandas: FTBFS on armhf and sparc: Bus error in test_append_frame_column_oriented
On 15/02/16 13:36, Yaroslav Halchenko wrote: Sorry about delay... quite often those Bus errors just go away on their own since aren't anything to be fixed in pandas, but this time it still persists with 0.17.1: https://buildd.debian.org/status/fetch.php?pkg=pandas=armhf=0.17.1-3=1450078009 test_append_frame_column_oriented (pandas.io.tests.test_pytables.TestHDFStore) ... Exception AttributeError: AttributeError("'File' object has no attribute '_node_manager'",) in ignored Bus error sparc is gone and seems to be not trying on sparc64 yet. I will look into fixing up my testing environment on sparc64 to see how current master is doing: http://nipy.bic.berkeley.edu/builders/pandas-py2.x-sid-sparc/builds/1806 Dear ARM people -- any clues on what could be a culprit here The most likely cause of "bus error" on armhf is an alignment issue with a pointer to a floating point type.
imx6 ethernet drops with sid kernel.
On one of the wandboard quads I use as an autobuilder for raspbian a btrfs filesystem got into a state where mount hung. After some googling the reccomended fix was a kernel upgrade so I upgraded to the kernel from sid. Linux bm-wb-01 4.3.0-1-armmp #1 SMP Debian 4.3.3-7 (2016-01-19) armv7l GNU/Linux However the board now seems to be having issues with ethernet dropping from time to time. The most recent time it happened I was on the serial console and saw the following messages. [ 9348.474418] fec 2188000.ethernet eth0: MDIO read timeout [ 9350.478804] fec 2188000.ethernet eth0: Link is Down Some googling shows reports of similar issues but i'm not sure if/how these apply to the Debian kernel.
Re: Squeeze support
On 28/11/15 13:06, Mikhail Kotelnikov wrote: Hi. I have Netgear ReadyNAS with OS based on Debian armel arch. When I try to update it to current version of squeeze all mirrors miss some deb packages. My sources.list: === cut === deb http://ftp.debian.org/debian/ squeeze main contrib non-free deb-src http://ftp.debian.org/debian/ squeeze main contrib non-free deb http://security.debian.org/ squeeze/updates main contrib non-free deb-src http://security.debian.org/ squeeze/updates main contrib non-free deb http://ftp.debian.org/debian/ squeeze-updates main contrib non-free deb-src http://ftp.debian.org/debian/ squeeze-updates main contrib non-free === cut === Also tried us.debian.org and de.debian.org. Use archive.debian.org https://lists.debian.org/debian-devel-announce/2015/08/msg9.html
Re: Mixing Fortran and C (on arm64, ppc64, ppc64el, mips64el and s390x)
On 25/09/15 21:57, Edmund Grimley Evans wrote: Does anyone here have any experience of mixing Fortran and C? A couple of (possibly) related bug reports: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=774618 https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=800040 Right. I don't use fortran but I took a look at the bug reports anyway. #if defined(__alpha__) || defined(__sparc64__) || defined(__x86_64__) || defined(__ia64__) I think you could just add the failing 64-bit architectures to that list: defined(__aarch64__) for arm64; see https://wiki.debian.org/ArchitectureSpecificsMemo for the others. That just seems like papering over the problem. (Note that this is, presumably, not supposed to be a list of all "64-bit architectures". The package seems to be working on s390x and ppc64, although they are "64-bit architectures" and not in the list I belive your conclusion is wrong. From the ppc64 build log (s390x was much the same). ┌──┐ │ Run tests for ruby2.1 from debian/ruby-tests.rb │ └──┘ RUBYLIB=/«PKGBUILDDIR»/debian/ruby-lapack/usr/lib/powerpc64-linux-gnu/ruby/vendor_ruby/2.1.0:/«PKGBUILDDIR»/debian/ruby-lapack/usr/lib/ruby/vendor_ruby:. ruby2.1 debian/ruby-tests.rb Loaded suite tests Started ** On entry to CGESDD parameter number 5 had an illegal value /usr/bin/ruby2.2 /usr/bin/gem2deb-test-runner On the other hand from the alpha build log I see (armel and i386 were much the same). ┌──┐ │ Run tests for ruby2.1 from debian/ruby-tests.rb │ └──┘ RUBYLIB=/«PKGBUILDDIR»/debian/ruby-lapack/usr/lib/alpha-linux-gnu/ruby/vendor_ruby/2.1.0:/«PKGBUILDDIR»/debian/ruby-lapack/usr/lib/ruby/vendor_ruby:. ruby2.1 debian/ruby-tests.rb Loaded suite tests Started ... . Finished in 0.232545348 seconds. -- 116 tests, 448 assertions, 0 failures, 0 errors, 0 pendings, 0 omissions, 0 notifications 100% passed -- 498.83 tests/s, 1926.51 assertions/s /usr/bin/ruby2.2 /usr/bin/gem2deb-test-runner So I would conclude that the code is broken on all 64-bit architectures not explicitly enumerated in the ifdef but the breakage manifests itself differently on big endian and little endian architectures and for whatever reason the manifestation seen on big endian architectures doesn't cause a build failure. Personally I would suggest changing the ifdef to #if defined(__alpha__) || defined(__sparc64__) || defined(__x86_64__) || defined(__ia64__) || defined(__LP64__) (you could remove the specific CPUs that are already there but I left them in just in case anyone is building with an old compiler) I would then suggest running test builds on at least one architecture from each category (e.g. at least one out of arm64/mips64el/ppc64el and at least one out of s390x/ppc64) and seeing if the situation improves.
Re: Bug#795331: iceweasel: backport/security builds (bundled libvpx) FTBFS on armhf
On 13/08/15 00:05, Simon McVittie wrote: As described onhttps://bugs.debian.org/795079, a rebuild of iceweasel/38.1.1esr-1 with ~deb9u1 appended to the version number, in a stretch environment, failed to build from source on armhf. Build log: https://people.debian.org/~jcristau/iceweasel_38.1.1est-1~deb9u1_armhf.log.txt I worked around this by using ~stretch instead of ~deb9u1 for the subsequent 38.2.0 upload. I suspect this might also be affecting 38.2.0esr-1~deb8u1 in the security archive. security.debian.org offers 38.2.0esr-1~deb8u1 for armel but not armhf, which supports this theory. However, security buildds don't publish their logs to avoid leaking details of embargoed packages Yeah, it sucks that there is no system in place to publish logs for security packages along with the packages themselves. The iceweasel packaging automatically adapts itself to be suitable for backports when it sees ~bpo or ~deb in the version number: yuck in particular, it uses a few bundled libraries, including libvpx, instead of the system equivalents. If I had re-generated control.in from control, it would also have dropped the corresponding build-dependencies, but for a t-p-u upload I didn't particularly want to do that. Instead of its upstream build system as used in src:libvpx, the bundled libvpx in iceweasel is built using media/libvpx/moz.build, which has this code: if CONFIG['VPX_ARM_ASM']: ... for f in SOURCES: if f.endswith('.c') and 'neon' in f: SOURCES[f].flags += ['-march=armv7-a', '-mthumb', '-mfloat-abi=softfp', '-mfpu=neon'] I assume this is present so that the NEON-optimized files will be built with the softfp ABI on ARM OABI (the deprecated arm architecture in Debian), which used a different ABI by default. Using a non-default ABI seems like a good way to break stuff and afaict few people care about OABI anymore. I doubt this was the reason. On ARM EABI softfloat (armel), -mfloat-abi=softfp is equivalent to the default, and is harmless. No it's not equivilent to the default. The default on armel is mfloat-abi=soft which disables the FPU completely. mfloat-abi=softfp uses the soft-float ABI but allows the compiler to use the FPU So if you are going to build an alternative code path/version of a binary for armel that uses the fpu (obviously there must be a non-fpu based codepath/binary as well as presense of a VFP FPU is not gauranteed) then you need to use -mfloat-abi=softfp. Unfortunately, on ARM EABI hardfloat (armhf), the use of -mfloat-abi=softfp causes glibc to take the softfloat code path ingnu/stubs.h: #if !defined __ARM_PCS_VFP # includegnu/stubs-soft.h #endif #if defined __ARM_PCS_VFP # includegnu/stubs-hard.h #endif In Debian,gnu/stubs-soft.h only exists on armel, whilegnu/stubs-hard.h only exists on armhf, so this fails: code that tries to use the soft float abi on debian armhf is going to be pretty broken, so it's probablly a good thing that it fails to build.
Bug#785770: RM: xserver-xorg-video-siliconmotion [arm64] -- RoP; FTBFS; out-of-date; uninstallable; quite possiblly never worked, little if any utility
Package: ftp.debian.org x-debbugs-cc: debian-arm@lists.debian.org,debia...@lists.debian.org A recent xorg update caused xserver-xorg-siliconmotion to fail to build with the following error. /bin/bash ../libtool --tag=CC --mode=compile gcc -std=gnu99 -DHAVE_CONFIG_H -I. -I../../src -I.. -D_FORTIFY_SOURCE=2 -fvisibility=hidden -I/usr/include/pixman-1 -I/usr/include/libdrm -I/usr/include/xorg -I/usr/include/X11/dri -Wall -Wpointer-arith -Wmissing-declarations -Wformat=2 -Wstrict-prototypes -Wmissing-prototypes -Wnested-externs -Wbad-function-cast -Wold-style-definition -Wdeclaration-after-statement -Wunused -Wuninitialized -Wshadow -Wmissing-noreturn -Wmissing-format-attribute -Wredundant-decls -Wlogical-op -Werror=implicit -Werror=nonnull -Werror=init-self -Werror=main -Werror=missing-braces -Werror=sequence-point -Werror=return-type -Werror=trigraphs -Werror=array-bounds -Werror=write-strings -Werror=address -Werror=int-to-pointer-cast -Werror=pointer-to-int-cast -fno-strict-aliasing -g -O2 -fstack-protector-strong -Wformat -Werror=format-security -c -o smi_501.lo ../../src/smi_501.c libtool: compile: gcc -std=gnu99 -DHAVE_CONFIG_H -I. -I../../src -I.. -D_FORTIFY_SOURCE=2 -fvisibility=hidden -I/usr/include/pixman-1 -I/usr/include/libdrm -I/usr/include/xorg -I/usr/include/X11/dri -Wall -Wpointer-arith -Wmissing-declarations -Wformat=2 -Wstrict-prototypes -Wmissing-prototypes -Wnested-externs -Wbad-function-cast -Wold-style-definition -Wdeclaration-after-statement -Wunused -Wuninitialized -Wshadow -Wmissing-noreturn -Wmissing-format-attribute -Wredundant-decls -Wlogical-op -Werror=implicit -Werror=nonnull -Werror=init-self -Werror=main -Werror=missing-braces -Werror=sequence-point -Werror=return-type -Werror=trigraphs -Werror=array-bounds -Werror=write-strings -Werror=address -Werror=int-to-pointer-cast -Werror=pointer-to-int-cast -fno-strict-aliasing -g -O2 -fstack-protector-strong -Wformat -Werror=format-security -c ../../src/smi_501.c -fPIC -DPIC -o .libs/smi_501.o In file included from ../../src/smi.h:317:0, from ../../src/smi_501.c:33: ../../src/regsmi.h: In function 'VGAIN8_INDEX': ../../src/regsmi.h:67:2: error: implicit declaration of function 'outb' [-Werror=implicit-function-declaration] outb(pSmi-PIOBase + indexPort, index); ^ ../../src/regsmi.h:67:2: warning: nested extern declaration of 'outb' [-Wnested-externs] ../../src/regsmi.h:68:2: error: implicit declaration of function 'inb' [-Werror=implicit-function-declaration] return(inb(pSmi-PIOBase + dataPort)); ^ ../../src/regsmi.h:68:2: warning: nested extern declaration of 'inb' [-Wnested-externs] cc1: some warnings being treated as errors Further investigation (see https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=785491) showed that the reason for the FTBFS was the removal of non-functional stub implementations of those functions from xorg. I belive the out of date binary should be removed from sid for the following reasons. 1: it's uninstallable with the xorg currently in sid/stretch 2: while I have not investigated throughly it seems very likely that the use of stub versions of those functions would have rendered the driver unusable. 3: the siliconmotion chips in question seem to be obsolete parts mostly used for onboard graphics in old systems (though I did find one industrial minipcie card that claimed to use a siliconmotion chip, dunno if it's one supported by this driver). It seems unlikely that one would end up in an arm64 system. If this removal request is actioned I will downgrade bug 785491 -- To UNSUBSCRIBE, email to debian-arm-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/555bda04.30...@p10link.net
arm64 porterboxes
In order to answer a query on the xserver-xorg-video-siliconmotion bug report I tried to log into asachi.debian.org and got plugwash@debian:~$ ssh asachi.debian.org ssh: connect to host asachi.debian.org port 22: No route to host plugwash@debian:~$ The traceroute would seem to indicate that the problem is at or near the server end. traceroute to asachi.debian.org (64.28.108.84), 30 hops max, 60 byte packets 1 192.168.1.1 (192.168.1.1) 12.045 ms 12.033 ms 12.016 ms 2 * * * 3 0278095e.bb.sky.com (2.120.9.94) 26.176 ms 26.178 ms 26.170 ms 4 ae51.edge6.London1.Level3.net (212.113.9.53) 26.160 ms 26.129 ms 26.117 ms 5 ae-0-11.bar1.Houston1.Level3.net (4.69.137.133) 115.703 ms 114.634 ms 115.686 ms 6 ONRAMP-ACCE.bar1.Houston1.Level3.net (4.34.204.186) 121.314 ms 120.535 ms 120.532 ms 7 216.109.208.18 (216.109.208.18) 131.140 ms 131.064 ms 131.052 ms 8 onr108-84.onr.com (64.28.108.84) 130.981 ms 130.990 ms 130.979 ms 9 onr108-84.onr.com (64.28.108.84) 3128.088 ms !H 3039.879 ms !H 3038.847 ms !H plugwash@debian:~$ So I then tried turfan I was able to log in but I got. Could not chdir to home directory /home/plugwash: No such file or directory And there was no info in the motd about chroots etc. Are things just handled manually on that box? -- To UNSUBSCRIBE, email to debian-arm-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/5557f284.7050...@p10link.net
Re: Guide to getting ported?
On 01/05/15 18:32, Paul Wise wrote: On Fri, May 1, 2015 at 6:07 PM, Wookey wrote: Sarting off by saying in this doc 'Host' is 'BUILD' and 'Target' is 'HOST' seems like a great way to add to the confusion :-) Fixed. I don't see the edit fixing it. did you forget to save? Assuming that the dpkg-architecture variables are likely to get a mention in any such discussion it seems sensible to just use the GNU (and thus Debian) DEB_BUILD_ARCH and DEB_HOST_ARCH and DEB_TARGET_ARCH terminology and explain that, rather than introducing a different one. Does DEB_TARGET_ARCH actually exist? Looks like it: https://wiki.debian.org/MultiarchCrossToolchainBuild#Multiarch_Cross_Toolchain_Build -- To UNSUBSCRIBE, email to debian-arm-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/554406d0.2020...@p10link.net
chromium on arm (was: About Jessie and missing packages)
On Mon, Mar 30, 2015 at 1:21 AM, Umut KARCI master.rep...@gmail.com mailto:master.rep...@gmail.com wrote: I'm very grateful for your work on raspbian, I've been using it since the first pi. But currently I'm having a problem. I've upgraded to Jessie lately. Since yesterday, I can't install chromium browser, I checked mirrors and main repo, source packages are there but no deb packages. Debian dropped arm support from their chromium packages because I couldn't get it to work when built with the toolchain Debian had at the time (this was in wheezy) and noone else was stepping up to fix it either. It was crashing and I was unable to get a usable backtrace from the crash. See https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=696909 for more information on the many issues we had trying to keep chromium working on armhf. raspbian wheezy continues to ship the last chromium package I managed to make work. We are not currently shipping a chromium package in raspbian jessie. I don't recall exactly why I removed it but I think that the old version of chromium was no longer compatible with new library versions or something like that. If there is to be a supported chromium package in raspbian going forward it needs to be dealt with on the Debian side first. Unfortunately chromium seems to be the package from hell. It looks like since that time that the Debian chromium maintainers have tried to reenable armhf support on two occasions (in versions 28.0.1500.71-1 and 29.0.1547.57-3+exp1) but in both cases the package failed to build. ccing the chromium maintainers to see if they have any comments. Are you recompiling jessie packages for better armv7 support? The whole point of raspbian is to provide support for armv6. If you want armv7 optimised packages use Debian armhf.
Re: Question about build box hardware
On 02/03/15 19:02, Lennart Sorensen wrote: Well I was seeing a ton of crashes building openjdk (it rarely survived it), and ceph, when running 3.14 linaro kernel on the arndale-octa. With 3.19 plain release kernel, I have built openjdk, ceph and others a number of times (even did 3 builds of openjdk in parallel to abuse it) and have had no crashes at all. So perhaps something in 3.14 was unstable (and could still be true in 3.16) for some are systems, that 3.19 has fixed. I don't know which kernel the jessie builders are currently running. Assuming you mean the raspbian jessie autobuilders they are currently running a 3.17 kernel from Debian experimental. I tried Debian 3.18 but had issues with udev not working. IIRC wheezy's version refused to start while jessie/sid's version claimed to start successfully but didn't actually populate /dev . -- To UNSUBSCRIBE, email to debian-arm-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/54f506fb.1090...@p10link.net
Re: Question about build box hardware
On 27/02/15 15:27, Lennart Sorensen wrote: I believe raspian has managed failly well with an army of imx53qsb systems, with 1GB ram and single core, but quite a lot of them. Both Debian armhf and raspbian used those boards successfully during the wheezy release cycles, they coped in sufficient quantities (I think debian had 7, raspbian has 8) but some builds (e.g. many webkit variants) spent literally days in swap hell and at least one (pypy) outright refused to build because of the low ram. The original model of the imx53 quickstart board had the problem that it needed a button push to turn it on, the later R model with a different PMIC boots at power on (yay) but will not supply power the USB port when running a Debian wheezy kernel (it did work when running the freescale kernel, I haven't tested with a post-wheezy Debian kernel). In raspbian we are currently still building wheezy updates on the imx53 boards (I was meaning to move it over to the new infrastructure alongside jessie but haven't got round to it) but are building jessie stuff on quad core imx6 machines with 2GB of ram. For stuff like webkit the difference is night and day. AIUI Debian stuck with the mx53 boards through much of the jessie cycle before replacing them fairly recently with some donated marvell boards (which I don't think are readilly available to us plebs), not sure if any of the IMX53 boards are still in use by Debian. When they started that was one of the cheapest and easiest to get systems with 1GB ram. And some of us other debian arm users were recommending them. :) Indeed, at the time the two main choices I was aware of that were reasonablly affordable and had 1GB of ram were the pandaboard and the imx53 quickstart board. I was told at the time by people I trusted that ubuntu were having big problems with USB relibaility on the pandaboard. I understand these issues were later traced to a kernel bug but by that time our descision was long made. Of course I would not reccomend either the IMX53 quickstart board or the pandaboard to someone starting now. The bang per buck is TERRIBLE by current standards. Now a cubox-i4 might not be a bad choice either. 4 core Cortex-A9 with 2GB ram and SATA (something the arndale-octa does not have, it only has onboard eMMC and a uSD slot, as well as a USB3 port). AIUI That is substantially the same as the wandboard quads and nitrogen6x we use for raspbian jessie, just in a different form factor. Those boards haven't been too bad in general with us. We have been having some crashes during iceweasel builds (strangely it very rarely crashes when building anything else but has a high probability of crashing when building iceweasel) but I have my suspiscions that those may be more down to btrfs than to our choice of board. Certainly the quad A15 is faster than the quad A9, so the arndale is probably still a faster compile system. I would expect you are right, USB3 makes me nervous not so much because of performance as because USB in general does not see to be a particulally dependable storage interface (for example I have USB3 hard drives from a major hard drive vendor where mearly touching the cable gently will cause a dropout, this tells me something about how seriously the vendors take reliability on their USB products). -- To UNSUBSCRIBE, email to debian-arm-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/54f3991b.4030...@p10link.net
Re: Expanding memory of an armel device?
On 27/02/15 00:26, Paul Wise wrote: On Thu, 2015-02-26 at 23:47 +, peter green wrote: Umm the odriod C1 and the raspberry Pi 2 model B (there is no raspberry pi 2 model A currently, this may or may not change in future) both have 1GB of ram. Bah, was mislead by the wiki pages. Fixed the FreedomBox one. The RPi2 Wikipedia page redirects to RPi and mediawiki didn't change the URL so I didn't notice. Could you followup on the list? Yeah, I meant to send my previous mail to the list but accidently sent it to you privately, i've sent it to the list now. Also they both lack SATA, so for NAS-type use i'd be looking more towards something like the cubieboards Good point, could you followup on the list? Especially if there any NAS form-factor or plug-computer form-factor devices with Cubie stuff. AIUI The cubietruck comes with plexiglass covers which can mount a 2.5 inch HDD. If you want full enclosure (the covers cover the top and bottom but not the sides) then those seem to be available too. -- To UNSUBSCRIBE, email to debian-arm-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/54efc148.8010...@p10link.net
Re: Expanding memory of an armel device?
On 26/02/15 13:29, Paul Wise wrote: There appear to be PCIe SSDs but I and others couldn't find any PCIe RAM, People have successfully set up swap on graphics cards in the past, I don't know how feasible that is with modern cards though (I guess they are now using a windowing system or so to avoid taking up too much address space). there appears to be a gap in the market. You can get FPGAs with PCIe and memory interfaces and i'm sure you could program one to make the memory visible on the PCIe bus but the cost would be dispropotionate to the benefit. To do this efficiently would really need a dedicated PCIe to DDR3 chip and I don't think there is enough of a market to support that. -- To UNSUBSCRIBE, email to debian-arm-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/54ef70c1.5060...@p10link.net
Re: Expanding memory of an armel device?
On 26/02/15 22:58, Paul Wise wrote: I think this is what it comes down to, any hardware method to increase the RAM of such an old machine is probably going to cost more than just upgrading to a more recent board like the RPi2 (256MB/512MB/1GB) or ODroid C1 (2GB). Umm the odriod C1 and the raspberry Pi 2 model B (there is no raspberry pi 2 model A currently, this may or may not change in future) both have 1GB of ram. Also they both lack SATA, so for NAS-type use i'd be looking more towards something like the cubieboards -- To UNSUBSCRIBE, email to debian-arm-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/54efbe18.5060...@p10link.net
Re: Question about build box hardware
On 27/02/15 00:44, Stephen Paul Weber wrote: Storage interfaces can also be an issue. USB is ubiquitous on arm boards but the implementations can be shaky and this has made me wary Do you need a lot of on-device storage all at once? With gigabit ethernet, having an on-LAN mirror + NAS for storing results might work as long as there's enough space to build the current task? Or, I may be missing something. You will want a few gigabytes of swap (2GB of ram may be enough to build stuff in less than a day, it's not enough to build stuff without swapping, IIRC I have 5GB of swap configured on the raspbian autobuilders though that may be overkill). You also need space for the build chroot and the build tree inside it which can easilly reach into the gigabytes (I looked at a libreoffice build and it was over 5GB, it wouldn't surprise me if there is some stuff that goes beyond 10GB though I can't name any offhand). Ideally the build chroot should also be restored to a clean snapshot after each build. In raspbian we use local HDDs with btrfs snapshots for this*, I think Debian uses LVM snapshots HDDs on some buildds and SSDs on others but i'm not positive on that. I would not want to put this kind of thing on SD cards, USB sticks, etc because of the heavy write load which these devices are simply not designed for. Putting it on some kind of networked storage (I've wondered about exporting ramdrives over NBD from a large-ram x86 box in the past and using them as swap on arm boards but i've never actually tried it) may be an option on boards that have fast network but I have not tried such setups. * Though if I had to start from scratch I suspect I would go for the (less space efficient) lvm snapshots instead. I suspect some of the build failures we get in raspbian are a result of our btrfs use. -- To UNSUBSCRIBE, email to debian-arm-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/54efc920.6070...@p10link.net
ccing bug reports to debian-arm (was: Bug#778898: FTBFS on arm64)
On 21/02/15 13:53, Chen Baozi wrote: Add dh-autoreconf to build-dep to fix FTBFS on arm64 I can understand ccing bug reports that are complex and need more eyes to debian-arm, I do it myself occasionally but I don't think it's nessacery or productive to cc debian-arm on trivial buildsystem fixes like this. -- To UNSUBSCRIBE, email to debian-arm-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/54e89b1c.30...@p10link.net
Re: Bug#764732: gcc-4.9: broken -O2 optimizations on armhf
Matthias Klose wrote: there exist several workarounds for it (lowering the optimization, using gcc-4.8, ...). Disabling stack protector also seems to result in a succesful compile (reducing it from strong to regular does not). -- To UNSUBSCRIBE, email to debian-arm-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/549cdaf0.4000...@p10link.net
Re: Bug#764732: gcc-4.9: broken -O2 optimizations on armhf
peter green wrote: Matthias Klose wrote: there exist several workarounds for it (lowering the optimization, using gcc-4.8, ...). Disabling stack protector also seems to result in a succesful compile (reducing it from strong to regular does not). And another workaround is to use -marm. -- To UNSUBSCRIBE, email to debian-arm-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/549cdc99.6010...@p10link.net
issues with experimental kernels on imx6 based boards.
I have some imx6 based boards which work as autobuilders for raspbian. I use btrfs for efficient snapshotting and have had some issues (crashes, weird filesystem errors) which I suspect are down to btrfs. As such i've been frequently upgradeing the kernel to the latest versions from experimental in the hope that the issues will be fixed upstream. There are a total of six boards. Five wandboard quads (one hosted at my flat, four hosted at bytemark), and one nitrogen6x, the wandboard quads are running a mostly wheezy system while the nitrogen6x is running a mostly jessie system. On all the systems the boot files and ext4 root filesystem is on a micro SD card while a hard drive is used for buildd related stuff (btrfs) and swap. The wandboards have a seperate ext2 boot partition while the nitrogen6x has /boot on the root partition. Currently there are (among others) the following kernels installed on the boards. linux-image-3.16-trunk-armmp version 3.16-1~exp1 linux-image-3.17-1-armmp version 3.17-1~exp1 linux-image-3.18.0-trunk-armmp version 3.18-1~exp1 linux-image-3.16-trunk-armmp seemed to work ok without any tweaking. linux-image-3.17-1-armmp seemed to work ok without any tweaking on the wandboard quads but it failed to boot with a timeout waiting for rootfs. I was able to get it to boot by doing lsmod | cut -d ' ' -f 1 /etc/initramfs-tools/modules (while running 3.16) and then rebuilding the initrd for 3.17, i'll try to investigate further whether this is reproducable with later versions of initramfs-tools and if-so what modules exactly are missing. linux-image-3.18.0-trunk-armmp, this failed on the nitrogen6x in the same way as 3.17, I haven't debugged that issue further yet. On the wandboard quad it booted but /dev was nearly empty and udev was failing to start, I tried upgrading udev first to the wheezy-backports version and then to the jessie version but while udev claimed to start successfully it still didn't seem to actually start managing /dev. Is anyone aware of any udev incompatibilities with 3.18? -- To UNSUBSCRIBE, email to debian-arm-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/54975d0d.1060...@p10link.net
swp on armv8 (Was: Haskell on arm needs help)
Joachim Breitner wrote: Hi, Am Mittwoch, den 03.12.2014, 23:02 +0100 schrieb Joachim Breitner: Trying to use it, but ghc fails to install in the schroot, and using just dd-schroot-cmd, I cannot debug this. Does installing ghc work properly for you? $ dd-schroot-cmd -c ghc apt-get install ghc Reading package lists... Building dependency tree... Reading state information... ghc is already the newest version. 0 upgraded, 0 newly installed, 0 to remove and 15 not upgraded. 1 not fully installed or removed. Conf ghc (7.6.3-20 Debian:unstable [armhf]) Do it for real [Y/n]: Reading package lists... Building dependency tree... Reading state information... ghc is already the newest version. 0 upgraded, 0 newly installed, 0 to remove and 15 not upgraded. 1 not fully installed or removed. After this operation, 0 B of additional disk space will be used. Download complete and in download only mode Reading package lists... Building dependency tree... Reading state information... ghc is already the newest version. 0 upgraded, 0 newly installed, 0 to remove and 15 not upgraded. 1 not fully installed or removed. After this operation, 0 B of additional disk space will be used. Setting up ghc (7.6.3-20) ... Illegal instruction Illegal instruction dpkg: error processing package ghc (--configure): subprocess installed post-installation script returned error exit status 132 Errors were encountered while processing: ghc E: Sub-process /usr/bin/dpkg returned an error code (1) Command apt-get --assume-yes -o Dpkg::Options::=--force-confnew install -- ghc exited with exit code 1. there is more things strage with this machine. The above was the sid schroot (presuambly arm64). I think the default chroot arch follows the userland arch of the outer system and the outer system on asachi is armhf. I can confirm that on asachi ghc fails to install in the armhf chroot, installs but fails to run in the armel chroot and installs and runs in the arm64 chroot. On the armel chroot on that machine I can install ghc, but cannot run it successfully: ~/ghc-7.8.20141119 $ ghc Illegal instruction ~/ghc-7.8.20141119 $ ghc --version Illegal instruction (This is calling the ghc from the package from unstable.) I vaugely remember something a while back about some deprecated 32-bit arm instructions needing kernel emulation on armv8 and that emulation not being implemented yet. After some fighting with gdb I got (sid_armel-dchroot)plugwash@asachi:~$ gdb /usr/lib/ghc/lib/ghc --snip gdb startup-- Reading symbols from /usr/lib/ghc/lib/ghc...(no debugging symbols found)...done. (gdb) run Starting program: /usr/lib/ghc/lib/ghc Cannot access memory at address 0x0 [Thread debugging using libthread_db enabled] Using host libthread_db library /lib/arm-linux-gnueabi/libthread_db.so.1. Program received signal SIGILL, Illegal instruction. 0x02921958 in ?? () (gdb) --snip fighting with gdb-- (gdb) disassemble 0x02921958,0x02921962 Dump of assembler code from 0x2921958 to 0x2921962: = 0x02921958: swp r3, r4, [r5] 0x0292195c: cmp r3, #0 0x02921960: beq 0x292197c End of assembler dump. (gdb) swp has been deprecated for a while, armhf binaries should really avoid using it but sadly other than runtime CPU detection i'm not sure there is much choice for armel. AIUI swp is already handled through kernel emulation on armv7 multiprocessor systems. There seem to be patches to port that emulation to arm64 but it doesn't appear they are in the kernel tree debian is using. Having 32-bit binaries break on armv8 systems due to lack of the swp instruction does not seem like a good thing so IMO we really want this in our kernels before release. Putting debian-arm back in cc and adding debian-kernel to cc to hopefully get some comments from people more knowlagable, Should I use a different machine to debug armel build failures? Or is ghc in unstable (and hence testing) that broken on armel? Theres a few armv7 porterboxes available though with lower resources than asachi abel.debian.org (I think this is the best resourced one) harris.debian.org ipa.debian.net (non-dsa) -- To UNSUBSCRIBE, email to debian-arm-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/547f9b7d.6080...@p10link.net
armv7 builds leaking into armel
I just filed a couple of bugs about packages whose build systems are building for armv7 on armel[1][2]. While doing so I also noticed that openmpi also seems to have similar problems but that case is more complex and I don't have the time/energy to investigate it right now. We have a patch in raspbian[3] but it needs some work to figure out what should be done in Debian. The two issues mentioned above were using armv7 cflags unconditionally on all arm architctures but i've also come across ones in the past that use armv7 cflags only if the system they are being built on is armv7 [4]. Since Debian recently switched to armv7 autobuilders for armel it is likely that Debian has latent issues of this type. For the most part armv7 binaries can be detected using readelf. It does suffer from some false positives (runtime cpu detection etc) and false negatives (jit compilers etc) but it should be sufficient to discover most issues. In Raspbian we run all packages we build through a check that unpacks them and runs readelf[5]. I try to push issues I detect this way back up to Debian but I sometimes forget and there are some less clear cut cases that I don't have the time to triage (for example with neon related ones there may or may not be runtime detection, in raspbian I just tend to disable neon completely but that may not be appropriate for Debian). I don't think Debian armel should be relying on Raspbian to find these issues. Anyone here who cares about armel fancy putting something together to check the armel archive for armv7 binaries? or are we just going to leave them arround until some user gets the illegal instruction error? [1] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=770782 [2] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=770793 [3] http://debdiffs.raspbian.org/main/o/openmpi/openmpi_1.6.5-8+rpi1.debdiff [4] https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=719355 [5] https://lists.debian.org/debian-arm/2013/05/msg00154.html -- To UNSUBSCRIBE, email to debian-arm-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/5472b18c.2050...@p10link.net
Haskell on arm needs help (was: hledger and hledger-web packages for raspbian)
oops, forgot to actually cc debian-arm Readers on debian-arm and raspbian-devel, see https://lists.debian.org/debian-haskell/2014/11/msg00024.html if you want more context. Joachim Breitner wrote: There was some support for ghci on arm, but it wasn’t working well and disabled it in Debian in 7.6.3: * Disable GHCi on arm, as it is severly borken (for more information see http://hackage.haskell.org/trac/ghc/ticket/7794) Maybe things have improved in 7.8 (currently in experimental). We could use some manpower dedicated to testing this, hunting down upstream patches and trying them out. Unfortunately, the current 7.8 package does not even build on arm: https://buildd.debian.org/status/package.php?p=ghcsuite=experimental and every developer trying to fix it gave up, mostly because the long build times make it so annoying. You might find asachi.debian.org useful, it's an APM x-gene mustang with 8 64-bit cores. Not sure of the clockspeed but i'm fairly sure it's high. The box has armel, armhf and arm64 chroots available for use. Maybe there are people out there in the Raspbian community who care for ghc (not just ghci!) on arm and would like to fix these problems before they migrate to Raspbian? You’d be welcome to join the Debian Haskell Group! Well it won't be me, I don't know haskell and I don't have the time/inclination to learn it. Sending this to debian-arm and raspbian-devel to see if anyone there can help out. -- To UNSUBSCRIBE, email to debian-arm-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/546fb852.4070...@p10link.net
arm64 box under $1K for developers in the USA
I just saw an advert on LWN which people here might find interesting. It looks like apm have reduced the price of their box to $1495 and are offering a developer special at $895. Still more than i'm personally willing to pay and it looks like the offer is US only (for non-us customers they point to avnet who still seem to be quoting a price of about $3K) https://www.apm.com/products/data-center/x-gene-family/x-c1-development-kits/x-c1-development-kit-basic/ -- To UNSUBSCRIBE, email to debian-arm-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/54681850.4030...@p10link.net
Re: State of arm64 in testing.
peter green wrote: gadmin-openvpn-client0.1.2-4unsatisfied dependency on openvpn gadmin-openvpn-client-dbg0.1.2-4unsatisfied dependency on openvpn gadmin-openvpn-server0.1.5-3.1 unsatisfied dependency on openvpn gadmin-openvpn-server-dbg 0.1.5-3.1unsatisfied dependency on openvpn openvpn-auth-ldap2.0.3-6.1unsatisfied dependency on openvpn (= 2) openvpn-auth-radius2.1-6+b1unsatisfied dependency on openvpn (= 2) openvpn-auth-radius-dbg2.1-6+b1unsatisfied dependency on openvpn (= 2) openvpn has been binnmu'd in tpu, hopefully it will show up as fixed after the next run. The binnmu has made it successfully into testing fixing these. gnome-core-devel1:3.8+8unsatisfied dependency on libseed-gtk3-dev (= 3.2) The release team have told me this will be fixed by dropping the depedendency Dependency has been dropped in sid and unblocked, waiting for it to migrate. tcos-core0.89.93unsatisfied dependency on xtightvncviewer tightvnc is fixed in sid, an unblock request has been filed with the release team. The fixed tightvnc has now migrated to testing making tcos-core installable. xfce4-goodies4.1unsatisfied dependency on xfce4-battery-plugin This has been fixed in sid but afaict an unblock request has not yet been filed, i'll probablly do so in a day or two if noone else gets there first. xfce4-battery-plugin migrated, unfortunately it turns out it wasn't the only thing making xfce4-goodies uninstallable. We now have xfce4-goodies4.1unsatisfied dependency on xfce4-notes-plugin I'm investigating this now and will file a bug later, hopefully with a patch. snort2.9.7.0-3unsatisfied dependency on libdaq0 snort-common-libraries2.9.7.0-3unsatisfied dependency on libdaq0 These have just appeared due to the migration of a new version of snort to testing, I suspect a TPU binnmu of daq is the appropriate response, again i'll probablly request it in a day or so if noone else does so first. I investigated this in more detail and it was messier than this and the mess wasn't really arm64 specific, see the thread starting with https://lists.debian.org/debian-release/2014/11/msg00339.html for more details. -- To UNSUBSCRIBE, email to debian-arm-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/546117a3.1010...@p10link.net
Re: State of arm64 in testing.
peter green wrote: peter green wrote: I've been looking at the list (https://qa.debian.org/dose/debcheck/testing_main/1413608403/arm64.html) of uninstallable packages in arm64 testing focussing on packages that are arch specific (since AIUI current de-facto policy is it's acceptable for arch all packages to be uninstallable but unacceptable for arch specific packages to be uninstallable). Time for an update. the stiaution has imrpoved massively but we still have some ways to go to get the list of uninstallable arch specific binaries down to 0 Getting very close now. The remaining arch specific entries on the dose page are (i'm pretty sure this is correct, unfortunately the dose pages don't allow filtering to arch specific packages making it difficult to be positive) gadmin-openvpn-client0.1.2-4unsatisfied dependency on openvpn gadmin-openvpn-client-dbg0.1.2-4unsatisfied dependency on openvpn gadmin-openvpn-server0.1.5-3.1unsatisfied dependency on openvpn gadmin-openvpn-server-dbg0.1.5-3.1unsatisfied dependency on openvpn openvpn-auth-ldap2.0.3-6.1unsatisfied dependency on openvpn (= 2) openvpn-auth-radius2.1-6+b1unsatisfied dependency on openvpn (= 2) openvpn-auth-radius-dbg2.1-6+b1unsatisfied dependency on openvpn (= 2) openvpn has been binnmu'd in tpu, hopefully it will show up as fixed after the next run. gnome-core-devel1:3.8+8unsatisfied dependency on libseed-gtk3-dev (= 3.2) The release team have told me this will be fixed by dropping the depedendency tcos-core0.89.93unsatisfied dependency on xtightvncviewer tightvnc is fixed in sid, an unblock request has been filed with the release team. xfce4-goodies4.1unsatisfied dependency on xfce4-battery-plugin This has been fixed in sid but afaict an unblock request has not yet been filed, i'll probablly do so in a day or two if noone else gets there first. snort2.9.7.0-3unsatisfied dependency on libdaq0 snort-common-libraries2.9.7.0-3unsatisfied dependency on libdaq0 These have just appeared due to the migration of a new version of snort to testing, I suspect a TPU binnmu of daq is the appropriate response, again i'll probablly request it in a day or so if noone else does so first. -- To UNSUBSCRIBE, email to debian-arm-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/54597619.9020...@p10link.net
Bug#767439: python-zmq FTBFS on official arm64 arm-arm autobuilders.
Package: python-zmq Version: 14.3.1-1 Severity: important User: debian-arm@lists.debian.org Usertags: arm64 x-debbugs-cc: debian-arm@lists.debian.org Python-zmq (versions 14.3.1-1 and 14.4.0-1) FTBFS on the official arm-arm arm64 autobuilders. The same versions succeeded on the debian-ports arm64 autobuilder turfan. The arm-linaro official arm64 autobuilders have never tried to build the package. The failure takes the form of a test failure Test connected monitoring socket. ... Assertion failed: pfd.revents POLLIN (signaler.cpp:193) Aborted E: pybuild pybuild:256: test: plugin distutils failed with: exit code=134: cd /«PKGBUILDDIR»/.pybuild/pythonX.Y-dbg_3.4/build; python3.4-dbg -m unittest discover -v dh_auto_test: pybuild --test -i python{version}-dbg -p 3.4 --dir . returned exit code 13 debian/rules:51: recipe for target 'override_dh_auto_test' failed Anyone have any thoughts? This failure is making one arch specific package and a number of arch all packages uninstallable in arm64 testing. -- To UNSUBSCRIBE, email to debian-arm-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/5452ef1a.2080...@p10link.net
Re: State of arm64 in testing.
peter green wrote: I've been looking at the list (https://qa.debian.org/dose/debcheck/testing_main/1413608403/arm64.html) of uninstallable packages in arm64 testing focussing on packages that are arch specific (since AIUI current de-facto policy is it's acceptable for arch all packages to be uninstallable but unacceptable for arch specific packages to be uninstallable). Time for an update. the stiaution has imrpoved massively but we still have some ways to go to get the list of uninstallable arch specific binaries down to 0 1: gcc-4.9/binutils: gcc-4.9 picked up a tighter binutils dependency on arm64 than on other architectures. Unfortunately this means it's not installable with the binutils currently in jessie. Binutils has not been migrating to jessie due to frequent uploads. Binutils has now finally migrated to testing :). 2: gcc-defaults: gcc-defaults has not been migrating to jessie for a while due to depdency issues on kfreebsd. It was binnmu'd successfully in arm64 tpu and that binnmu migrated (I suspect it was forced but I don't know for sure) to testing but unfortunately several of the binary packages built from gcc-defaults are not binnmu safe and are hence uninstallable. This has apparently just migrated, it will take some time for the dose pages to update to reflect that. 3: many dependencies sitting in needs-build: hopefully the extra autobuilders that have just come online will help with that. This is no longer an issue. 4: notify-python, wookey bugrepped this about 6 days ago, probablly about time to turn that bugrep into a NMU. hippo-canvas: also seems to need a NMU for config.sub/guess issues. dvdauthor: also seems to need a NMU for config.sub/guess issues. Theese are fixed in sid but waiting for migration to testing. Other packages that need to be dealt with cython: blocked from migrating to testing by build failures on armel (testsuite failure) and mips (timeout). May be a candiate for a TPU binnmu later (probablly not the right time to do it yet) seed: blocked from migration to testing by build failures on mips, powerpc and s390x) These two are still outstanding tightvnc: also seems to need a NMU for config.sub/guess issues. As wookey pointed out I misdiagnosed this before, i'm now taking a more detailed look. New issues/concerns gnat-4.9: needs to migrate to testing and for that it needs to be built on mips, it's been sitting in needs-build on mips for a worryingly long time. vinagre: waiting for migration to testing plplot: appears to be blocked from migrating to testing by gnat-4.9 (see above) haskell: I think this may just need some time to settle after binutils got in, needs keeping an eye on over the next few days. magics++: weird issue relating to -data package but a new versions migrating from unstable soon, hopefully that will fix things xfce4-battery-plugin: needs a small fix for arm64, sledge posted a bugrep a few days ago, probablly wait a couple more days then NMU if there is no maintainer response. -- To UNSUBSCRIBE, email to debian-arm-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/545018f6.7070...@p10link.net