Re: [gentoo-user] Radeon navi cards and opencl
On Tuesday, 31 August 2021 18:21:48 BST Matt Connell (Gmail) wrote: > On Tue, 2021-08-31 at 16:51 +0100, Peter Humphrey wrote: > > AMD have published Linux drivers on their website, but in RPM and .deb > > formats. I tried one anyway, but installation stops immediately with > > 'unsupported OS'. Not surprising, of course, but is there any way of > > porting an archive full of .debs so that it can be used in Gentoo? > > Lots of portage ebuild files make use of vendor-published .deb / .rpm > files, manipulating them to install files in the Gentoo-style way. > > One of the more straightforward and yet complete examples I've seen is > the Microsoft Teams (net-im/teams) package[1] but there are certainly > others to use for examples. > > If you wanted, you could try making an ebuild for the deb files from > AMD, whether only for your own needs or to possibly share with others. That's a huge undertaking in this case. Beyond me, I deem. > 1:https://gitweb.gentoo.org/repo/gentoo.git/tree/net-im/teams/teams-1.4.00.1 > 3653-r1.ebuild The thing is, there's already an ebuild for dev-libs/amdgpu-pro-opencl, but since I upgraded the display card the resulting code hasn't worked. BOINC reports 'No usable GPUs found' and sources on the Web say that neither of the two drivers works. Impasse. Am I going to have to replace the old display card and write off the expense? Looks like it. -- Regards, Peter.
Re: [gentoo-user] rescue cd for zfs 2.1 or thereabouts
On Tue, 31 Aug 2021 20:44:50 +0100, Michael wrote: > > systemd-boot and refind both support everything on EFI. I am pretty > > sure GRUB does too, but I have no reason to use GRUB with EFI. My > > setup on this box is /boot on FAT32 and / (and everything else) on > > btrfs. I've also used the same setup with ZFS. > > Any boot option on a UEFI MoBo requires an 'EFI System Partition' > (ESP), formatted as VFAT. The UEFI firmware boot loader will > list/load/run any *.efi software stored in the ESP compatible with the > UEFI API, whether this is a boot loader, a kernel with an EFI stub, or > some .efi diagnostic application. > > As long as your boot loader of choice, or kernel image and any initrd > contains the requisite fs drivers, there will be no problem mounting > and accessing whatever root fs needs to be accessed. > > GRUB contains a number of ZFS modules to do this job (zfscrypt.mod, > zfsinfo.mod, zfs.mod) - not sure about the other boot managers. > > Typical GRUB installations have /boot/efi mounted on the ESP, with the > grubx64.efi image on it, while the rest of the files, vmlinuz symlinks, > etc. are on the root partition. > > Please beware, I have not used zfs to date, only btrfs, so the above > merely reflects my understanding rather than in depth experience of the > difficulty in managing such a setup. I find it simpler to make /boot a FAT partition, then /boot/efi is the ESP and all boot-related files are on the same filesystem. Like you, I have only used this with btrfs (and ext4 on LVM). I do use ZFS but that system is not EFI. -- Neil Bothwick Theory is when you know everything, but nothing works. Reality is when everything works, but you don't know why. However, usually theory and reality are mixed together : Nothing works, and nobody knows why not. pgpimEf06G9h9.pgp Description: OpenPGP digital signature
Re: [gentoo-user] rescue cd for zfs 2.1 or thereabouts
On 2021.08.31 16:20, Michael wrote: [snip] Is there some particular need to resize boot? Is there even a need to have a pool for it? It seems to me having a stand alone simple VFAT partition for / boot which doubles up as ESP with a /boot/EFI subdirectory will do the job, although I understand not all use cases are as simplistic as what I envisage here. My only need to resize my boot/ESP partition was when I decided to increase the number of kernel/initrd combinations I wanted to keep around at one time, and ran out of space. Jack
Re: [gentoo-user] rescue cd for zfs 2.1 or thereabouts
On Tuesday, 31 August 2021 20:54:14 BST Rich Freeman wrote: > On Tue, Aug 31, 2021 at 3:44 PM Michael wrote: > > Please beware, I have not used zfs to date, only btrfs, so the above > > merely > > reflects my understanding rather than in depth experience of the > > difficulty in managing such a setup. > > To save you digging through the thread, the issue with zfs is that it > adds new features over time, and grub isn't necessarily compatible > with all of them. You can control which features are enabled on-disk > for compatibility, but grub doesn't do a great job documenting which > features are/aren't supported in any particular version. So, it is a > bit of a guessing game. > > It has been pointed out that there are various guides online, but: > > 1. They don't all say the exact same thing. > 2. They aren't official upstream docs. > 3. They rarely specify what version of grub they're talking about. > > The typical solution is to either use very conservative settings for > your root partition (which isn't ideal from a zfs standpoint), or have > a separate /boot pool which means that you don't have to encumber the > rest of the system with whatever grub's limitations might be. Then > you just never update that partition and it shouldn't break. That > basically is no different than just having /boot on ext4 or vfat or > whatever. With this solution you also can't just freely resize /boot > the way you could if it were part of the pool. Yes, I understood this much, although I haven't red the various online guides you refer to above. Is there some particular need to resize boot? Is there even a need to have a pool for it? It seems to me having a stand alone simple VFAT partition for / boot which doubles up as ESP with a /boot/EFI subdirectory will do the job, although I understand not all use cases are as simplistic as what I envisage here. signature.asc Description: This is a digitally signed message part.
Re: [gentoo-user] rescue cd for zfs 2.1 or thereabouts
On Tue, Aug 31, 2021 at 3:44 PM Michael wrote: > > Please beware, I have not used zfs to date, only btrfs, so the above merely > reflects my understanding rather than in depth experience of the difficulty in > managing such a setup. To save you digging through the thread, the issue with zfs is that it adds new features over time, and grub isn't necessarily compatible with all of them. You can control which features are enabled on-disk for compatibility, but grub doesn't do a great job documenting which features are/aren't supported in any particular version. So, it is a bit of a guessing game. It has been pointed out that there are various guides online, but: 1. They don't all say the exact same thing. 2. They aren't official upstream docs. 3. They rarely specify what version of grub they're talking about. The typical solution is to either use very conservative settings for your root partition (which isn't ideal from a zfs standpoint), or have a separate /boot pool which means that you don't have to encumber the rest of the system with whatever grub's limitations might be. Then you just never update that partition and it shouldn't break. That basically is no different than just having /boot on ext4 or vfat or whatever. With this solution you also can't just freely resize /boot the way you could if it were part of the pool. -- Rich
Re: [gentoo-user] rescue cd for zfs 2.1 or thereabouts
On Tuesday, 31 August 2021 20:21:58 BST Neil Bothwick wrote: > On Tue, 31 Aug 2021 14:21:35 -0400, Rich Freeman wrote: > > > If you are multibooting frequently and getting into the UEFI boot > > > menu to change the boot order or running efibootmgr is too much > > > hassle, then a 3rd party boot manager will be useful. Your choice of > > > GRUB, rEFInd, systemd-boot, syslinux, EFI executable image will be > > > installed and loaded/run by the UEFI firwmare from the ESP, with > > > which in turn you will select and load your desired OS. > > > > So, which (if any) of these options supports either: > > > > 1. An EFI partition plus /boot on zfs (with no limitations on pool > > config, ie it can be a root pool). > > 2. An EFI partition that contains everything. > > > > If I want to use grub+EFI with a zfs root it sounds like I'd need TWO > > boot partitions - an EFI partition (FAT32), and a /boot partition > > (anything, but if ZFS it needs to have controlled features). That > > seems even more messy than what I'm doing now. > > systemd-boot and refind both support everything on EFI. I am pretty sure > GRUB does too, but I have no reason to use GRUB with EFI. My setup on > this box is /boot on FAT32 and / (and everything else) on btrfs. I've > also used the same setup with ZFS. Any boot option on a UEFI MoBo requires an 'EFI System Partition' (ESP), formatted as VFAT. The UEFI firmware boot loader will list/load/run any *.efi software stored in the ESP compatible with the UEFI API, whether this is a boot loader, a kernel with an EFI stub, or some .efi diagnostic application. As long as your boot loader of choice, or kernel image and any initrd contains the requisite fs drivers, there will be no problem mounting and accessing whatever root fs needs to be accessed. GRUB contains a number of ZFS modules to do this job (zfscrypt.mod, zfsinfo.mod, zfs.mod) - not sure about the other boot managers. Typical GRUB installations have /boot/efi mounted on the ESP, with the grubx64.efi image on it, while the rest of the files, vmlinuz symlinks, etc. are on the root partition. Please beware, I have not used zfs to date, only btrfs, so the above merely reflects my understanding rather than in depth experience of the difficulty in managing such a setup. signature.asc Description: This is a digitally signed message part.
Re: [gentoo-user] rescue cd for zfs 2.1 or thereabouts
On Tue, 31 Aug 2021 14:21:35 -0400, Rich Freeman wrote: > > If you are multibooting frequently and getting into the UEFI boot > > menu to change the boot order or running efibootmgr is too much > > hassle, then a 3rd party boot manager will be useful. Your choice of > > GRUB, rEFInd, systemd-boot, syslinux, EFI executable image will be > > installed and loaded/run by the UEFI firwmare from the ESP, with > > which in turn you will select and load your desired OS. > > > > So, which (if any) of these options supports either: > > 1. An EFI partition plus /boot on zfs (with no limitations on pool > config, ie it can be a root pool). > 2. An EFI partition that contains everything. > > If I want to use grub+EFI with a zfs root it sounds like I'd need TWO > boot partitions - an EFI partition (FAT32), and a /boot partition > (anything, but if ZFS it needs to have controlled features). That > seems even more messy than what I'm doing now. systemd-boot and refind both support everything on EFI. I am pretty sure GRUB does too, but I have no reason to use GRUB with EFI. My setup on this box is /boot on FAT32 and / (and everything else) on btrfs. I've also used the same setup with ZFS. -- Neil Bothwick If a book about failures doesn't sell, is it a success? pgpcYZfW2Hspk.pgp Description: OpenPGP digital signature
Re: [gentoo-user] rescue cd for zfs 2.1 or thereabouts
On Tue, Aug 31, 2021 at 1:51 PM Michael wrote: > > > > > In a small nutshell, you have a small EFI+boot partition, set to type > > 'EFI System' and formatted FAT32, then tell grub to use it as an EFI > > directory when calling grub-install. > > In simple(r) systems where you only boot the same OS you can instead use the > kernel's EFI stub to get the UEFI firmware to load the latest OS kernel > directly from the ESP, without a 3rd party boot manager: > > https://wiki.gentoo.org/wiki/EFI_stub > > You'll use the efibootmgr to manage the kernel images stored on ESP, or your > UEFI configuration menu if it has this functionality. > > https://wiki.gentoo.org/wiki/Efibootmgr > > > If you are multibooting frequently and getting into the UEFI boot menu to > change the boot order or running efibootmgr is too much hassle, then a 3rd > party boot manager will be useful. Your choice of GRUB, rEFInd, systemd-boot, > syslinux, EFI executable image will be installed and loaded/run by the UEFI > firwmare from the ESP, with which in turn you will select and load your > desired OS. > So, which (if any) of these options supports either: 1. An EFI partition plus /boot on zfs (with no limitations on pool config, ie it can be a root pool). 2. An EFI partition that contains everything. If I want to use grub+EFI with a zfs root it sounds like I'd need TWO boot partitions - an EFI partition (FAT32), and a /boot partition (anything, but if ZFS it needs to have controlled features). That seems even more messy than what I'm doing now. -- Rich
[gentoo-user] Thunderbird w/ Owl plugin for OWA?
Does anybody have any experience using Thunderbird with the Owl plugin for connecting to Exchange OWA? https://addons.thunderbird.net/en-US/thunderbird/addon/owl-for-exchange/ I've been thinking it might be a better option than hiri. https://www.hiri.com/ I've been using hiri for a couple years now, but development and support seemed to have stopped several years ago. It still works as well as it always did, but there are years-old bugs that don't look like they're ever going to get fixed, and there are some rather basic features that are missing or don't work. On the pro side, the calendar function does sync up and work pretty well. The basic mail functionality works good enough about 80% of the time. But, every month or two an e-mail thread gets broken and you have to delete the account and start from scratch. The editor is also lacking basic operations like indent, quote, code block, font/color, insert image, and a few others. Oh, and there's no "undo" in the editor. About once a day, my emacs muscle memory will kick in and I'll hit ctrl-a to move the beginning of the line and then start typing something. Poof, the entire draft is gone with no way to recover. I've found that a viable work-around for anything at all long or complex is to edit the message using markdown-mode in emacs, preview the message in a browser, and then cut/paste it into hiri's editor window. But, the OWA web UI works just as well for that... -- Grant Edwards grant.b.edwardsYow! This ASEXUAL PIG at really BOILS my BLOOD gmail.com... He's so ... so ... URGENT!!
Re: [gentoo-user] rescue cd for zfs 2.1 or thereabouts
On Tuesday, 31 August 2021 15:33:07 BST Matt Connell (Gmail) wrote: > On Tue, 2021-08-31 at 10:04 -0400, Rich Freeman wrote: > > What I probably should get around to is grokking EFI+linux. I'm not > > sure what the cleanest solution for that is these days - I've never > > actually set up EFI on linux, mostly because I'm not sure what the > > best practice is. > > I can't speak to best practices, and I'm a long way from an expert on > EFI, but the the instructions on the Gentoo handbook pages on disk > partitioning + grub setup worked as expected for me. > > In a small nutshell, you have a small EFI+boot partition, set to type > 'EFI System' and formatted FAT32, then tell grub to use it as an EFI > directory when calling grub-install. In simple(r) systems where you only boot the same OS you can instead use the kernel's EFI stub to get the UEFI firmware to load the latest OS kernel directly from the ESP, without a 3rd party boot manager: https://wiki.gentoo.org/wiki/EFI_stub You'll use the efibootmgr to manage the kernel images stored on ESP, or your UEFI configuration menu if it has this functionality. https://wiki.gentoo.org/wiki/Efibootmgr If you are multibooting frequently and getting into the UEFI boot menu to change the boot order or running efibootmgr is too much hassle, then a 3rd party boot manager will be useful. Your choice of GRUB, rEFInd, systemd-boot, syslinux, EFI executable image will be installed and loaded/run by the UEFI firwmare from the ESP, with which in turn you will select and load your desired OS. https://wiki.gentoo.org/wiki/GRUB2 http://www.rodsbooks.com/refind/ https://wiki.gentoo.org/wiki/Systemd-boot https://wiki.syslinux.org/wiki/index.php?title=Install#UEFI signature.asc Description: This is a digitally signed message part.
Re: [gentoo-user] Xscreensaver fails to start : solved
210831 Arve Barsnes wrote: > On Tue, 31 Aug 2021 at 16:27, Philip Webb wrote: >> I just updated to the latest Xscreensaver-6.01-r1. > Did you mean -r3 ? Yes, a typo. >> I emerged it with USE flag 'elogind' & don't use Systemd. >> 'which xscreensaver-systemd' finds nothing >> nor can I see any helpful messages in a file in /var/log . >> I have set the binary's permissions to SUID : >> root:526 /usr/bin> ls -l xscreensaver >> -rws--s--x 1 root root 51880 Aug 31 07:18 xscreensaver > Not sure if this is helpful, but I have neither elogind or systemd > in USE for this package, and it seems to work fine. It seems to be working correctly now, tho' only time will tell. I believe I needed to reboot to get the new version set up properly & instead I initially tried a "setuid root", which used to be required. Now apparently, it is strickly forbidden (grimace). Anyway, after a remerge & reboot & no "setuid", it got going as it should. -- ,, SUPPORT ___//___, Philip Webb ELECTRIC /] [] [] [] [] []| Cities Centre, University of Toronto TRANSIT`-O--O---' purslowatchassdotutorontodotca
Re: [gentoo-user] Laptop internal speakers no longer working after recent updates
On Tuesday, 31 August 2021 12:08:04 BST Alexander Puchmayr wrote: > Am Dienstag, 31. August 2021, 00:18:43 CEST schrieb Michael: > > If the alsa drivers are not compiled as modules, the above file would not > > have any effect. Anyway, let's try this in /etc/asound.conf: > > > > defaults.pcm.card 1 > > defaults.pcm.device 0 > > defaults.ctl.card 1 > > > > On a reboot your Generic_1 analogue card should be available and > > recognised > > as the default audio device. You may need to unmute it, via pactl or > > kmix. > > Sorry, didn't change anything. > > I doubt that the problem is wrong default settings of alsa. > > I run pulseaudio -vvv and the output was interesting: > > Pci:00/:00:08.1/:07:00.1/sound/card0 > and > pci:00/:00:08.1/:07:00.6/sound/card1 > > Where the latter one is the one that is not used by pulseaudio. > > Both report "UCM available for card HD-Audio Generic" > Note: the card name "HD-Audio Generic" is identical, and this is reported by > alsa-libs, as far as I could see from the code. Your *card* names according to your 'aplay -l' output are/were: card 0: Generic card 1: Generic_1 card 3: Headset You can re-check this with: aplay -l | awk -F \: '/,/{print $2}' | awk '{print $1}' | uniq Another way to discover all card name(s) including unused cards is by the output of: cat /sys/class/sound/card*/id The alsa-ucm function involves creating use case alsa mixer profiles, similar to pulseaudio profiles and will work even without pulseaudio running. If a UCM configuration file exists for a card, then pulseaudio will ignore built-in profiles and will generate a profile based on the UCM config file. Take a look at: /usr/share/alsa/ucm2/README.md and for various mixer profiles look under /usr/share/alsa/ucm2/ However, my usage of pulseaudio has been cursory and don't know much about its auto-configuration. In any case, I suspect the alsa-ucm output is only relevant in highlighting the common codec name, as you confirm below. > [...] > > Alsa-info.sh reveals further info: > !!Soundcards recognised by ALSA > !!- > > 0 [Generic]: HDA-Intel - HD-Audio Generic > HD-Audio Generic at 0xfd3c8000 irq 91 > 1 [Generic_1 ]: HDA-Intel - HD-Audio Generic > HD-Audio Generic at 0xfd3c irq 92 > 2 [acp]: acp - acp > acp > > To me it looks like as if pulseaudio is quering card0, getting the name "HD- > Audio Generic", finding the HDMI channels; then it tries to read card1, > gets also "HD-Audio Generic" as name and hence the same channels as for > card0. > > I have no idea how to fix this. > > Cheers > Alex As I understand it, "HD-Audio" is the kernel driver (CONFIG_SND_HDA=m) and "Generic" is the generic codec parser (CONFIG_SND_HDA_GENERIC=m) used by the snd-hda-intel module (CONFIG_SND_HDA_INTEL=m) - unless a specific model codec is (also) configured for a card, e.g. in my case I have CONFIG_SND_HDA_CODEC_CONEXANT=m If in your recent system update/upgrade you did not change your kernel, or the available options of any audio modules under /etc/modprobe.d/ then the drivers were always configured so and therefore the problem you experience now is unlikely to be caused by the generic codec parser. Someone more knowledgeable in pulseaudio should chime in, assuming this problem is being caused by pulseaudio. :-/ signature.asc Description: This is a digitally signed message part.
Re: [gentoo-user] Radeon navi cards and opencl
On Tue, 2021-08-31 at 16:51 +0100, Peter Humphrey wrote: > AMD have published Linux drivers on their website, but in RPM and .deb > formats. I tried one anyway, but installation stops immediately with > 'unsupported OS'. Not surprising, of course, but is there any way of porting > and archive full of .debs so that it can be used in Gentoo? Lots of portage ebuild files make use of vendor-published .deb / .rpm files, manipulating them to install files in the Gentoo-style way. One of the more straightforward and yet complete examples I've seen is the Microsoft Teams (net-im/teams) package[1] but there are certainly others to use for examples. If you wanted, you could try making an ebuild for the deb files from AMD, whether only for your own needs or to possibly share with others. 1:https://gitweb.gentoo.org/repo/gentoo.git/tree/net-im/teams/teams-1.4.00.13653-r1.ebuild
Re: [gentoo-user] Radeon navi cards and opencl
>On Monday, 30 August 2021 16:08:27 BST Peter Humphrey wrote: >> Hello list, >> >> Does anyone know whether AMD intend to make their navi cards usable with >> opencl under Linux? I have a Radeon Pro W5500, which is a navi 14 card, and >> from what I can see neither rocm-opencl nor amdgpu-pro-opencl can drive it. >> >> Rather spoils the point of upgrading to it. > >AMD have published Linux drivers on their website, but in RPM and .deb >formats. I tried one anyway, but installation stops immediately with >'unsupported OS'. Not surprising, of course, but is there any way of porting >and archive full of .debs so that it can be used in Gentoo? app-arch/deb2targz app-arch/rpm2targz perhaps? And you get to keep all the pieces if the expected APIs do not match the actual APIs. :) DaveF > >-- >Regards, >Peter. > > > >
Re: [gentoo-user] Radeon navi cards and opencl
On Monday, 30 August 2021 16:08:27 BST Peter Humphrey wrote: > Hello list, > > Does anyone know whether AMD intend to make their navi cards usable with > opencl under Linux? I have a Radeon Pro W5500, which is a navi 14 card, and > from what I can see neither rocm-opencl nor amdgpu-pro-opencl can drive it. > > Rather spoils the point of upgrading to it. AMD have published Linux drivers on their website, but in RPM and .deb formats. I tried one anyway, but installation stops immediately with 'unsupported OS'. Not surprising, of course, but is there any way of porting and archive full of .debs so that it can be used in Gentoo? -- Regards, Peter.
Re: [gentoo-user] rescue cd for zfs 2.1 or thereabouts
On Tue, 31 Aug 2021 09:33:06 -0400, John Covici wrote: > or if there is a way to add zfs to the sysresc cd, I might do > that. There is, I posted a link last week. >> You can also add additional modules to sysrescd, so it may be easier to >> stick with that and add ZFS to it. >> >> Looking at https://www.system-rescue.org/Modules/ it looks like you just >> install them after booting the live USB then run cowpacman2srm to put >> everything you installed onto a file that you add to the USB stick. See >> the section titled "Creating SRM modules out of pacman packages". -- Neil Bothwick Snacktrek, n.: The peculiar habit, when searching for a snack, of constantly returning to the refrigerator in hopes that something new will have materialized. pgpIK6tUhqTAQ.pgp Description: OpenPGP digital signature
Re: [gentoo-user] Xscreensaver fails to start
On Tue, 31 Aug 2021 at 16:27, Philip Webb wrote: > > I just updated to the latest Xscreensaver-6.01-r1. Did you mean -r3? I've been on that for over a month, and the only other version available to me is 5.45-r4. > I emerged it with USE flag 'elogind' & don't use Systemd. > 'which xscreensaver-systemd' finds nothing > nor can I see any helpful messages in a file in /var/log . > I have set the binary's permissions to SUID : > > root:526 /usr/bin> ls -l xscreensaver > -rws--s--x 1 root root 51880 Aug 31 07:18 xscreensaver > > Can anyone offer advice ? Not sure if this is helpful, but I have neither elogind or systemd in USE for this package, and it seems to work fine. Regards, Arve
Re: [gentoo-user] rescue cd for zfs 2.1 or thereabouts
On Tue, 2021-08-31 at 10:04 -0400, Rich Freeman wrote: > What I probably should get around to is grokking EFI+linux. I'm not > sure what the cleanest solution for that is these days - I've never > actually set up EFI on linux, mostly because I'm not sure what the > best practice is. I can't speak to best practices, and I'm a long way from an expert on EFI, but the the instructions on the Gentoo handbook pages on disk partitioning + grub setup worked as expected for me. In a small nutshell, you have a small EFI+boot partition, set to type 'EFI System' and formatted FAT32, then tell grub to use it as an EFI directory when calling grub-install.
[gentoo-user] Xscreensaver fails to start
I just updated to the latest Xscreensaver-6.01-r1. If it starts at reboot, individual savers crash with a screen box saying ' crashed with Status 1'. If I stop the daemon, then restart it from CLI, I get : xscreensaver-systemd: 10:15:26: user bus connection failed: No such file or directory xscreensaver: 10:15:26: pid 15832: xscreensaver-systemd exited unexpectedly with status 1 I emerged it with USE flag 'elogind' & don't use Systemd. 'which xscreensaver-systemd' finds nothing nor can I see any helpful messages in a file in /var/log . I have set the binary's permissions to SUID : root:526 /usr/bin> ls -l xscreensaver -rws--s--x 1 root root 51880 Aug 31 07:18 xscreensaver Can anyone offer advice ? -- ,, SUPPORT ___//___, Philip Webb ELECTRIC /] [] [] [] [] []| Cities Centre, University of Toronto TRANSIT`-O--O---' purslowatchassdotutorontodotca
Re: [gentoo-user] rescue cd for zfs 2.1 or thereabouts
On Tue, Aug 31, 2021 at 9:33 AM John Covici wrote: > > Well, I never use a boot pool, I boot with ext4 and just do the root > on zfs. But, I was more interested in some external media, so I am > looking at that Linux Recovery file system to see if I can get that > working, or if there is a way to add zfs to the sysresc cd, I might do > that. > Yeah, if I had to use a separate boot partition of any sort, I probably wouldn't make it zfs. There really isn't any benefit from using it for that. It might be FAT32 if I'm using EFI/etc, or it might be ext4. If I'm concerned about mirroring for something like that I could just rsync it once a day to zfs. We're talking about a tiny amount of data that doesn't change much, and if you use a week-old kernel after a restore it isn't the end of the world. What I probably should get around to is grokking EFI+linux. I'm not sure what the cleanest solution for that is these days - I've never actually set up EFI on linux, mostly because I'm not sure what the best practice is. I know at a high level that a few options exist. I want a final experience that is closer to grub2 than syslinux/lilo/etc, to use an analogy (not saying it should actually use grub). -- Rich
Re: [gentoo-user] rescue cd for zfs 2.1 or thereabouts
On Tue, 31 Aug 2021 07:11:53 -0400, Robert David wrote: > > Hi John, > > my approach is to have EFI partition with staticly compiled grub, alpine > linux rescue system and kernel with tiny initramfs for zfs root. > > It works really well, only thing you need to consider is when upgrading > root pool, you will not be able to boot to previous BE with old zfs. > Without upgrading the pool, the transition is just easy as recompiling > new kernel and upgrading the zfs userspace tools. > > For the alpine I use script to put a new version on /boot > https://github.com/robertek/root-scripts/blob/master/alpine_recovery_update > > and having grub entry: > > menuentry "Alpine linux recovery" { > linux /boot/vmlinuz-lts modules=loop,squashfs,sd-mod,nvme quiet > nomodeset > initrd /boot/initramfs-lts > } > > The alpine extended version contains zfs modules, so you only need to > "apk add zfs" and then modprobe zfs. > > The extended version is little bit bigger, but I'm fine to live with 1G efi > partition. > > Robert. > > On Monday, August 23, 2021 10:15:10 AM CEST John Covici wrote: > > Hi. I have been using 5.4 lts kernels for a while, but it seems I > > need to change to 5.10 lts -- even Debian is now using 5.10, so it > > seems time to do this. > > > > Now, the problem is that I am using zfs and will not give it up, and > > the version I have been using 0.8.6 is no longer supported in 5.10 > > versions of the kernel. So, I need a newer version of zfs and a > > rescue cd in case I get into trouble. Sysresc seems to no longer be > > compatible withgentoo linux, so what is available? I could use gentoo > > catalyst to make something -- I have done that in the past, but its > > quite a bit of work and I would prefer if there were something > > available I could use out of the box. > > > > Thanks in advance for any suggestions. > > > > > Well, I never use a boot pool, I boot with ext4 and just do the root on zfs. But, I was more interested in some external media, so I am looking at that Linux Recovery file system to see if I can get that working, or if there is a way to add zfs to the sysresc cd, I might do that. -- Your life is like a penny. You're going to lose it. The question is: How do you spend it? John Covici wb2una cov...@ccs.covici.com
[gentoo-user] Re: Motif compile fails
On 2021-08-28, Philip Webb wrote: > As part of updating to the latest stable Xscreensaver, > I tried to emerge its requirement Motif, which failed with this message : [...] > /usr/lib/gcc/x86_64-pc-linux-gnu/10.2.0/../../../../x86_64-pc-linux-gnu/bin/ld: > /usr/lib/gcc/x86_64-pc-linux-gnu/10.2.0/../../../../lib64/libfl.a(libmain.o): > relocation R_X86_64_PC32 against symbol `exit@@GLIBC_2.2.5' can not be > used when making a PIE object; recompile with -fPIE > /usr/lib/gcc/x86_64-pc-linux-gnu/10.2.0/../../../../x86_64-pc-linux-gnu/bin/ld: > final link failed: bad value > collect2: error: ld returned 1 exit status > make: *** [Makefile:502: wmluiltok] Error 1 > make: Leaving directory > '/var/tmp/portage/portage/x11-libs/motif-2.3.8-r2/work/motif-2.3.8-abi_x86_64.amd64/tools/wml' > * ERROR: x11-libs/motif-2.3.8-r2::gentoo failed (compile phase): > * emake failed > > -- end of Portage output -- > > This looks as if the problem is with Gcc flags, which is rather technical. > > Has anyone else run into this ? Does anyone have suggestions ? As the error involves libfl.a, from sys-devel/flex, I'd try rebuilding that package. See bug 583842, https://bugs.gentoo.org/show_bug.cgi?id=583842 -- Nuno Silva
Re: [gentoo-user] Re: Motif compile fails
210831 Nuno Silva wrote: > On 2021-08-28, Philip Webb wrote: >> As part of updating to the latest stable Xscreensaver, >> I tried to emerge its requirement Motif, which failed with this message : >[...] >> /usr/lib/gcc/x86_64-pc-linux-gnu/10.2.0/../../../../x86_64-pc-linux-gnu/bin/ld: >> /usr/lib/gcc/x86_64-pc-linux-gnu/10.2.0/../../../../lib64/libfl.a(libmain.o): >> relocation R_X86_64_PC32 against symbol `exit@@GLIBC_2.2.5' can not be >> used when making a PIE object; recompile with -fPIE >> /usr/lib/gcc/x86_64-pc-linux-gnu/10.2.0/../../../../x86_64-pc-linux-gnu/bin/ld: >> final link failed: bad value >> collect2: error: ld returned 1 exit status >> make: *** [Makefile:502: wmluiltok] Error 1 >> make: Leaving directory >> '/var/tmp/portage/portage/x11-libs/motif-2.3.8-r2/work/motif-2.3.8-abi_x86_64.amd64/tools/wml' >> * ERROR: x11-libs/motif-2.3.8-r2::gentoo failed (compile phase): >> * emake failed >> -- end of Portage output -- > As the error involves libfl.a, from sys-devel/flex, > I'd try rebuilding that package. > See bug https://bugs.gentoo.org/show_bug.cgi?id=583842 Thanks a lot : I was afraid no-one would reply. Yes, that worked & I've gone on to emerge the latest Xscreensaver-6.01-r3 , which has restored 'xscreensaver-settings' (was '-demo'), so I can finally suppress a couple of really ugly screensavers ! I did check Bugs for 'motif', but the bug above dates from 2016, so it didn't turn up in my search. -- ,, SUPPORT ___//___, Philip Webb ELECTRIC /] [] [] [] [] []| Cities Centre, University of Toronto TRANSIT`-O--O---' purslowatchassdotutorontodotca
Re: [gentoo-user] rescue cd for zfs 2.1 or thereabouts
Hi John, my approach is to have EFI partition with staticly compiled grub, alpine linux rescue system and kernel with tiny initramfs for zfs root. It works really well, only thing you need to consider is when upgrading root pool, you will not be able to boot to previous BE with old zfs. Without upgrading the pool, the transition is just easy as recompiling new kernel and upgrading the zfs userspace tools. For the alpine I use script to put a new version on /boot https://github.com/robertek/root-scripts/blob/master/alpine_recovery_update and having grub entry: menuentry "Alpine linux recovery" { linux /boot/vmlinuz-lts modules=loop,squashfs,sd-mod,nvme quiet nomodeset initrd /boot/initramfs-lts } The alpine extended version contains zfs modules, so you only need to "apk add zfs" and then modprobe zfs. The extended version is little bit bigger, but I'm fine to live with 1G efi partition. Robert. On Monday, August 23, 2021 10:15:10 AM CEST John Covici wrote: > Hi. I have been using 5.4 lts kernels for a while, but it seems I > need to change to 5.10 lts -- even Debian is now using 5.10, so it > seems time to do this. > > Now, the problem is that I am using zfs and will not give it up, and > the version I have been using 0.8.6 is no longer supported in 5.10 > versions of the kernel. So, I need a newer version of zfs and a > rescue cd in case I get into trouble. Sysresc seems to no longer be > compatible withgentoo linux, so what is available? I could use gentoo > catalyst to make something -- I have done that in the past, but its > quite a bit of work and I would prefer if there were something > available I could use out of the box. > > Thanks in advance for any suggestions.
Re: [gentoo-user] Laptop internal speakers no longer working after recent updates
Am Dienstag, 31. August 2021, 00:18:43 CEST schrieb Michael: > If the alsa drivers are not compiled as modules, the above file would not > have any effect. Anyway, let's try this in /etc/asound.conf: > > defaults.pcm.card 1 > defaults.pcm.device 0 > defaults.ctl.card 1 > > On a reboot your Generic_1 analogue card should be available and recognised > as the default audio device. You may need to unmute it, via pactl or kmix. Sorry, didn't change anything. I doubt that the problem is wrong default settings of alsa. I run pulseaudio -vvv and the output was interesting: Pci:00/:00:08.1/:07:00.1/sound/card0 and pci:00/:00:08.1/:07:00.6/sound/card1 Where the latter one is the one that is not used by pulseaudio. Both report "UCM available for card HD-Audio Generic" Note: the card name "HD-Audio Generic" is identical, and this is reported by alsa-libs, as far as I could see from the code. Then it finds HDMI1-3 for both cards. *** card0 *** [...] D: [pulseaudio] module-udev-detect.c: /devices/ pci:00/:00:08.1/:07:00.1/sound/card0 is busy: no D: [pulseaudio] module-udev-detect.c: Loading module-alsa-card with arguments 'device_id="0" name="pci-_07_00.1" card_name="alsa_card.pci-_07_00.1" namereg_fail=false tsched=yes fixed_latency_range=no ignore_dB=no deferred_volume=yes use_ucm=yes avoid_resampling=no card_properties="module- udev-detect.discovered=1"' D: [pulseaudio] reserve-wrap.c: Unable to contact D-Bus session bus: org.freedesktop.DBus.Error.NotSupported: Unable to autolaunch a dbus-daemon without a $DISPLAY for X11 I: [pulseaudio] alsa-ucm.c: UCM available for card HD-Audio Generic I: [pulseaudio] alsa-ucm.c: Set UCM verb to HiFi D: [pulseaudio] alsa-ucm.c: Got TQ for verb HiFi: HiFi D: [pulseaudio] alsa-ucm.c: Got PlaybackPCM for device HDMI3: hw:Generic,8 D: [pulseaudio] alsa-ucm.c: Got PlaybackPriority for device HDMI3: 1300 D: [pulseaudio] alsa-ucm.c: Got JackControl for device HDMI3: HDMI/DP,pcm=8 Jack W: [pulseaudio] alsa-ucm.c: UCM file does not specify 'PlaybackChannels' or 'CaptureChannels'for device HDMI3, assuming stereo duplex. D: [pulseaudio] alsa-ucm.c: No _conflictingdevs for device HDMI3 D: [pulseaudio] alsa-ucm.c: No _supporteddevs for device HDMI3 D: [pulseaudio] alsa-ucm.c: Got PlaybackPCM for device HDMI2: hw:Generic,7 D: [pulseaudio] alsa-ucm.c: Got PlaybackPriority for device HDMI2: 1200 D: [pulseaudio] alsa-ucm.c: Got JackControl for device HDMI2: HDMI/DP,pcm=7 Jack W: [pulseaudio] alsa-ucm.c: UCM file does not specify 'PlaybackChannels' or 'CaptureChannels'for device HDMI2, assuming stereo duplex. D: [pulseaudio] alsa-ucm.c: No _conflictingdevs for device HDMI2 D: [pulseaudio] alsa-ucm.c: No _supporteddevs for device HDMI2 D: [pulseaudio] alsa-ucm.c: Got PlaybackPCM for device HDMI1: hw:Generic,3 D: [pulseaudio] alsa-ucm.c: Got PlaybackPriority for device HDMI1: 1100 D: [pulseaudio] alsa-ucm.c: Got JackControl for device HDMI1: HDMI/DP,pcm=3 Jack W: [pulseaudio] alsa-ucm.c: UCM file does not specify 'PlaybackChannels' or 'CaptureChannels'for device HDMI1, assuming stereo duplex. D: [pulseaudio] alsa-ucm.c: No _conflictingdevs for device HDMI1 D: [pulseaudio] alsa-ucm.c: No _supporteddevs for device HDMI1 [...] *** card1 *** (the one that is ignored by pulseaudio) [...] D: [pulseaudio] module-udev-detect.c: /devices/ pci:00/:00:08.1/:07:00.6/sound/card1 is busy: no D: [pulseaudio] module-udev-detect.c: Loading module-alsa-card with arguments 'device_id="1" name="pci-_07_00.6" card_name="alsa_card.pci-_07_00.6" namereg_fail=false tsched=yes fixed_latency_range=no ignore_dB=no deferred_volume=yes use_ucm=yes avoid_resampling=no card_properties="module- udev-detect.discovered=1"' D: [pulseaudio] reserve-wrap.c: Unable to contact D-Bus session bus: org.freedesktop.DBus.Error.NotSupported: Unable to autolaunch a dbus-daemon without a $DISPLAY for X11 I: [pulseaudio] alsa-ucm.c: UCM available for card HD-Audio Generic I: [pulseaudio] alsa-ucm.c: Set UCM verb to HiFi D: [pulseaudio] alsa-ucm.c: Got TQ for verb HiFi: HiFi D: [pulseaudio] alsa-ucm.c: Got PlaybackPCM for device HDMI3: hw:Generic,8 D: [pulseaudio] alsa-ucm.c: Got PlaybackPriority for device HDMI3: 1300 D: [pulseaudio] alsa-ucm.c: Got JackControl for device HDMI3: HDMI/DP,pcm=8 Jack W: [pulseaudio] alsa-ucm.c: UCM file does not specify 'PlaybackChannels' or 'CaptureChannels'for device HDMI3, assuming stereo duplex. D: [pulseaudio] alsa-ucm.c: No _conflictingdevs for device HDMI3 D: [pulseaudio] alsa-ucm.c: No _supporteddevs for device HDMI3 D: [pulseaudio] alsa-ucm.c: Got PlaybackPCM for device HDMI2: hw:Generic,7 D: [pulseaudio] alsa-ucm.c: Got PlaybackPriority for device HDMI2: 1200 D: [pulseaudio] alsa-ucm.c: Got JackControl for device HDMI2: HDMI/DP,pcm=7 Jack W: [pulseaudio] alsa-ucm.c: UCM file does not specify 'PlaybackChannels' or 'CaptureChannels'for device HDMI2, assuming stereo duplex. D: [pulseaudio] alsa-uc
Re: [gentoo-user] how to run freerdp
On 8/31/21 9:45 AM, n952162 wrote: So, now that I've emerged xfreerdb, is there a reason it's better than rdesktop? Oh, I guess I know: USB support (etc.). VirtualBox offers that, but only from their Oracle proprietary code.
Re: [gentoo-user] how to run freerdp
On 8/31/21 8:36 AM, n952162 wrote: On 8/30/21 9:29 PM, tastytea wrote: On 2021-08-30 21:10+0200 n952162 wrote: I just emerged freerdp-2.3.2 without issue, but there's no executable except /usr/bin/winpr-*, which I think just serve to set up an authentication system. There's no freerdp, xfreerdp, xfreerdp-server, or anything. What am I not thinking about? I have the same version and it installed /usr/bin/xfreerdp. I guess you are missing a useflag? I have X alsa cups ffmpeg gstreamer jpeg pulseaudio usb xinerama xv enabled. Kind regards, tastytea Ah ... I guess I'll never learn. That did the trick, thank you. Now, though, I'm battling a different problem: [08:42:12:233] [14625:14626] [ERROR][com.freerdp.core.transport] - BIO_read returned a system error 11: Resource temporarily unavailable [08:42:12:233] [14625:14626] [ERROR][com.freerdp.core] - transport_read_layer:freerdp_set_last_error_ex ERRCONNECT_CONNECT_TRANSPORT_FAILED [0x0002000D] It reaches the other side, because if I change the port, using some arbitrary port, I get: [08:24:20:275] [14524:14525] [ERROR][com.freerdp.core.transport] - BIO_should_retry returned a system error 32: Broken pipe Comparing tcpdump traces of xfreerdp and rdesktop shows, I think, based on the TCP flags, the same failure: 09:37:58.222556 IP client.54938 > host.ms-wbt-server: Flags [*P*.], seq 1:46, ack 1, win 502, options [nop,nop,TS val 2653789238 ecr 2608962080], length 45 09:37:58.222713 IP host.ms-wbt-server > client.54938: Flags [.], ack 46, win 509, options [nop,nop,TS val 2608962080 ecr 2653789238], length 0 09:37:58.222822 IP host.ms-wbt-server > client.54938: Flags [*F*.], seq 1, ack 46, win 509, options [nop,nop,TS val 2608962080 ecr 2653789238], length 0 09:37:58.222841 IP host.ms-wbt-server > client.54938: Flags [*R*.], seq 2, ack 46, win 509, options [nop,nop,TS val 2608962080 ecr 2653789238], length 0 That would be PSH, FIN, RST? The port on the host only appears when I enable the remote console on the vm in VirtualBox. The authentication is External: VRDE: enabled (Address 0.0.0.0, Ports 3389, MultiConn: off, ReuseSingleConn: off, Authentication type: external) VRDE port: 3389 So, I guess this isn't an xfreerdp issue... So, now that I've emerged xfreerdb, is there a reason it's better than rdesktop?