Re: release status update
On Thu, Jul 29, 2004 at 03:11:11PM -0400, Joey Hess wrote: I've submitted what I hope will be the final debian-installer build to the autobuilders, but it seems that recent dependency breakage in newt will cause some of the builds to fail. I hope I can ask the autobuilders to retry, but this will likely delay things for a few days. At this point it's probably too late to get any changes into udebs in the initrds, and there are probably only a few days left to make changes to other udebs. Any changes will be synced to testing on a per-udeb basis again until we release, so you have to give me a very good reason to put in a change. Please consider including the base-installer 1.01 and the nobootloader 0.17 .udebs i uploaded today. The 1.01 base-installer include a nicer way to handle powerpc kernels with builtin initrd, and also makes sure that kernel upgraded after the install will be functional. The nobootloader change include giving exacter booting instructions for the powerpc/chrp_pegasos (a thing which could be cloned for other arches without boot loader but with known boot procedures), as well as add the logic for the generic case. Ideally, a string change would be needed to tell the user about the partition in which to find the kernel in the generic case, but this can wait for after the beta5/rc1 release, but it would look nicer. The string change is not a sever one though, just adding on partition ${BOOT} in the boot instruction to know where the kernel is. Friendly, Sven Luther -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
discover on powerbook [was: Re: release status update]
Il dom, 2004-04-25 alle 00:13, Petter Reinholdtsen ha scritto: [Giuseppe Sacco] A second problem I understand only now is that the powerpc discover we have in testing now is broken: every time I switch on my laptop (powerbook g4, 15, 867Mhz) it hangs when discover run and probe for parallel port devices. This means that I have to press the keyboard combination to reset the machine. I believe this is fixed in the version of discover1 that went into sarge last night. Discover was instructed to not probe paralell and serial ports at boot time, but there was a bug making it ignore this instruction. Please test if this is fixed or not, and add info to bug #237603 (and possibly #214790). Now I am upgrading the machine and it is downloading the new discover 2.0.4-3 with discover-data 2.2004.04.09-1, so this might be changed. I will let you know. I do not know if this is fixed in version 2 of discover. I suspect it is not. the upgrade was from testing 2004-04-17 to 2004-04-24 and discover changed from 1.5-2 to 2.0.4-3. The problem seems fixed, but I only tried once. I will do some more tests and I'll update the bug reports. Bye, Giuseppe -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: release status update
Steve Langasek wrote: - qlogicisp driver doesn't work right with many (most?) SCSI cards; bug #238593 et al. - both the de4x5 and the tulip drivers fail to work with many DECChip-based ethernet cards in the 2.4.2x series alpha kernels; bug #228654. Both of these sound like reasonable candidates for errata if they're not fixed, and not release blockers. Do you want them in the errata? It also seems fairly certain that the netboot images don't work on alpha (unless they were fixed while I wasn't looking). I don't seem to have a way to test these anyway, as telling SRM to boot from ethernet generates no network traffic here. Right, I see that's marked as being nonfunctonal on the ports status page. -- see shy jo signature.asc Description: Digital signature
Re: release status update
- both the de4x5 and the tulip drivers fail to work with many DECChip-based ethernet cards in the 2.4.2x series alpha kernels; bug #228654. Both of these sound like reasonable candidates for errata if they're not fixed, and not release blockers. Do you want them in the errata? does this bug affect the onboard tulip ethernet that comes with most of the alpha hardware? if so, its a real bad bug. elijah -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: release status update
On Sat, Apr 24, 2004 at 12:54:18PM -0400, Joey Hess wrote: Steve Langasek wrote: - qlogicisp driver doesn't work right with many (most?) SCSI cards; bug #238593 et al. - both the de4x5 and the tulip drivers fail to work with many DECChip-based ethernet cards in the 2.4.2x series alpha kernels; bug #228654. Both of these sound like reasonable candidates for errata if they're not fixed, and not release blockers. Do you want them in the errata? Yes, please. -- Steve Langasek postmodern programmer signature.asc Description: Digital signature
Re: release status update
On Sat, Apr 24, 2004 at 12:16:44PM -0500, elijah wright wrote: - both the de4x5 and the tulip drivers fail to work with many DECChip-based ethernet cards in the 2.4.2x series alpha kernels; bug #228654. Both of these sound like reasonable candidates for errata if they're not fixed, and not release blockers. Do you want them in the errata? does this bug affect the onboard tulip ethernet that comes with most of the alpha hardware? if so, its a real bad bug. It affects many (I don't know if we can say most) onboard tulip chips that came with alphas. There are actually a wide range of chips covered by the tulip and de4x5 drivers that were distributed by Digital; I don't know yet if there's a clear pattern of which chips are affected, or if the problem has more to do with the associated hardware (such as PCI configuration and chipsets). It is a very bad bug, but I don't know that it's feasible to do much about it in the 2.4 kernel series in time for sarge. Cheers, -- Steve Langasek postmodern programmer signature.asc Description: Digital signature
Re: release status update
Il gio, 2004-04-22 alle 18:55, Joey Hess ha scritto: [...] - powerpc: need list of working subarches (oldworld?) I would like to have powerpc release only after having kernel-image-2.4.25-powerpc-* version 2.4.25-8 in testing because of #245012. A second problem I understand only now is that the powerpc discover we have in testing now is broken: every time I switch on my laptop (powerbook g4, 15, 867Mhz) it hangs when discover run and probe for parallel port devices. This means that I have to press the keyboard combination to reset the machine. The problem appear *only* when the machine is powered on, and it's not appearing when I reboot it. When it hangs, the keyboard isn't working and the led (the only one this machine has) is on. Now I am upgrading the machine and it is downloading the new discover 2.0.4-3 with discover-data 2.2004.04.09-1, so this might be changed. I will let you know. Bye, Giuseppe -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: release status update
[Giuseppe Sacco] A second problem I understand only now is that the powerpc discover we have in testing now is broken: every time I switch on my laptop (powerbook g4, 15, 867Mhz) it hangs when discover run and probe for parallel port devices. This means that I have to press the keyboard combination to reset the machine. I believe this is fixed in the version of discover1 that went into sarge last night. Discover was instructed to not probe paralell and serial ports at boot time, but there was a bug making it ignore this instruction. Please test if this is fixed or not, and add info to bug #237603 (and possibly #214790). Now I am upgrading the machine and it is downloading the new discover 2.0.4-3 with discover-data 2.2004.04.09-1, so this might be changed. I will let you know. I do not know if this is fixed in version 2 of discover. I suspect it is not. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: release status update
Giuseppe Sacco wrote: I would like to have powerpc release only after having kernel-image-2.4.25-powerpc-* version 2.4.25-8 in testing because of #245012. If I'm expanding yuor wildcard correctly, they seem to go in tonight/tomorrow. Anyway, you have several days to get them in, but this does not feel like a release blocker to me if they do not make it. -- see shy jo signature.asc Description: Digital signature
Re: release status update
* Joey Hess [EMAIL PROTECTED] [2004-04-22 22:15]: was loaded since it has a different sub-architecture (however, iirc, delo-installer is also loaded on Cobalt, so I assume sub-arch handling is broken). Argh, not again. libdebian-installer 0.22 went into the archive today with some fixes for subarch handling, in particular #243692. Was that version on your image? No, I had 0.21.1. I'll test with 0.22 tonight; thanks for pointing this out. -- Martin Michlmayr [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: release status update
* tbm [EMAIL PROTECTED] [2004-04-23 01:43]: - arm: ? arm/netwinder most definitely works. It worked recently and there is It works; the only problem I had was that the wrong kernel was installed. For some reason, debian-installer/kernel/subarchitecture was not set. Unfortunately, I cannot quite figure out why. (I just tried on mipsel and it definitely gets set there.) -- Martin Michlmayr [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: release status update
Martin Michlmayr wrote: * tbm [EMAIL PROTECTED] [2004-04-23 01:43]: - arm: ? arm/netwinder most definitely works. It worked recently and there is It works; the only problem I had was that the wrong kernel was installed. For some reason, debian-installer/kernel/subarchitecture was not set. Unfortunately, I cannot quite figure out why. (I just tried on mipsel and it definitely gets set there.) Does archdetect show the correct output for netwinder? Thiemo -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: release status update
* Thiemo Seufer [EMAIL PROTECTED] [2004-04-23 16:32]: Does archdetect show the correct output for netwinder? Yes. -- Martin Michlmayr [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: release status update
* Joey Hess [EMAIL PROTECTED] [2004-04-22 22:15]: Argh, not again. libdebian-installer 0.22 went into the archive today with some fixes for subarch handling, in particular #243692. Was that version on your image? I tried it with 0.22 now and still see the same... it's quite strange. Just for reference: colo-installer is for mipsel/cobalt while delo-installer is for mipsel/r3k-kn02 and mipsel/r4k-kn04. On mipsel/r3k-kn02, it installs colo-installer and delo-installer, but on mipsel/cobalt it only installs colo-installer. Looking at the logs, I get this on mipsel/r3k-kn02 (egrep -n (colo|delo)-install syslog): 396:Apr 23 18:30:52 (none) user.debug anna[825]: DEBUG: install delo-installer, priority = standard 452:Apr 23 18:30:53 (none) user.debug anna[825]: DEBUG: search for package resolving kernel-installer, dependency from delo-installer 453:Apr 23 18:30:53 (none) user.debug anna[825]: DEBUG: install di-utils-mapdevfs, dependency from delo-installer 482:Apr 23 18:30:53 (none) user.debug anna[825]: DEBUG: install colo-installer, dependency from prebaseconfig 498:Apr 23 18:30:53 (none) user.debug anna[825]: DEBUG: search for package resolving kernel-installer, dependency from colo-installer 499:Apr 23 18:30:53 (none) user.debug anna[825]: DEBUG: install di-utils-mapdevfs, dependency from colo-installer 566:Apr 23 18:34:30 (none) user.debug main-menu[145]: DEBUG: search for package resolving kernel-installer, dependency from colo-installer 569:Apr 23 18:34:30 (none) user.info main-menu[145]: INFO: Falling back to the package description for colo-installer ... On mipsel/cobalt, I get: 417:Jan 1 00:01:09 (none) user.debug anna[924]: DEBUG: install colo-installer, priority = standard 487:Jan 1 00:01:09 (none) user.debug anna[924]: DEBUG: search for package resolving kernel-installer, dependency from colo-installer 488:Jan 1 00:01:09 (none) user.debug anna[924]: DEBUG: install di-utils-mapdevfs, dependency from colo-installer 531:Jan 1 00:01:09 (none) user.debug anna[924]: DEBUG: search for package resolving kernel-installer, dependency from colo-installer 532:Jan 1 00:01:09 (none) user.debug anna[924]: DEBUG: install di-utils-mapdevfs, dependency from colo-installer 596:Jan 1 00:02:26 (none) user.debug main-menu[143]: DEBUG: search for package resolving kernel-installer, dependency from colo-installer 599:Jan 1 00:02:26 (none) user.info main-menu[143]: INFO: Falling back to the package description for colo-installer 667:Jan 1 00:02:44 (none) user.debug main-menu[143]: DEBUG: search for package resolving kernel-installer, dependency from colo-installer ... AFAICT, the packages are done properly: Package: delo-installer Version: 0.003 Section: debian-installer Priority: standard Architecture: mipsel Depends: cdebconf-udeb, kernel-installer, di-utils-mapdevfs Provides: bootable-system Installed-Size: 44 Maintainer: Debian Install System Team [EMAIL PROTECTED] Description: Install Delo on a hard disk This udeb will install Delo on the hard disk. installer-menu-item: 73 subarchitecture: r3k-kn02 r4k-kn04 and: Package: colo-installer Version: 0.02 Section: debian-installer Priority: standard Architecture: mipsel Depends: cdebconf-udeb, kernel-installer, di-utils-mapdevfs Provides: bootable-system Installed-Size: 24 Maintainer: Debian Install System Team [EMAIL PROTECTED] Description: Install the Cobalt boot loader on your disk This udeb will install the Cobalt loader colo on your hard disk. installer-menu-item: 73 subarchitecture: cobalt Any idea what's going on? -- Martin Michlmayr [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: release status update
On Thu, Apr 22, 2004 at 12:55:06PM -0400, Joey Hess wrote: I'm trying to track status of various architectures at the top of TODO, here is what I have so far. - ia64: There is a devpts/sysvinit/glibc file conflict of some sort that breaks debootstrap. Need a bug #. (Might be related to #230857) This is bug #238963, which also affects alpha; it has been (hopefully) fixed by the recent -12 upload of glibc, which is currently in unstable but not in testing. - alpha: still SCSI cd problems? I've just seen that this is fixed in version 2.4.25-3 of the kernel-image-2.4.25-1-generic package. I can do a rebuild of the d-i kernel package this evening (~3hrs), or if anyone else is willing and able to get this uploaded sooner, feel free. -- Steve Langasek postmodern programmer signature.asc Description: Digital signature
Re: release status update
On Thu, Apr 22, 2004 at 12:55:06PM -0400, Joey Hess wrote: - m68k: need list of working subarches mac and amiga are fully supported. I hope to have support for the rest in the next couple of weeks or so, but if they aren't ready for release, so be it. -- Stephen R. Marenka If life's not fun, you're not doing it right! [EMAIL PROTECTED] signature.asc Description: Digital signature
Re: release status update
* Joey Hess [EMAIL PROTECTED] [2004-04-22 12:55]: - mipsel: looks good I just did a successful installation on a DECstation 5000/133 via the serial console. Everything worked fine, apart from one thing: both colo-installer and delo-installer got loaded, and colo-installer was executed instead of delo-installer. I have no idea why colo-installer was loaded since it has a different sub-architecture (however, iirc, delo-installer is also loaded on Cobalt, so I assume sub-arch handling is broken). p2 said that bogl crashes when he tries a DECstation install with a video card, but it worked for Karsten recently. I asked Karsten to see if it works for him. In any case, d-i definitely works on DECstation via the serial console and is ready for beta4. That is, mipsel/r3k-kn02 and mipsel/r4k-kn04 (DECstation with R3000 and R4000 CPUs). mipsel/cobalt also works. The only problem is that you (currently) need an existing Linux installation in order to boot d-i. This will be fixed after beta4 (I'm waiting for the upstream author of colo to implement net booting). Apart from this, Cobalt installations work. -- Martin Michlmayr [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: release status update
* Joey Hess [EMAIL PROTECTED] [2004-04-22 12:55]: I'm trying to track status of various architectures at the top of TODO, here is what I have so far. - arm: ? arm/netwinder most definitely works. It worked recently and there is afaict nothing that broke it in the meantime. arm/bast and arm/riscstation probably work as well. I hope I can do a test on arm/netwinder tomorrow to confirm. -- Martin Michlmayr [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: release status update
Martin Michlmayr wrote: I just did a successful installation on a DECstation 5000/133 via the serial console. Everything worked fine, apart from one thing: both colo-installer and delo-installer got loaded, and colo-installer was executed instead of delo-installer. I have no idea why colo-installer was loaded since it has a different sub-architecture (however, iirc, delo-installer is also loaded on Cobalt, so I assume sub-arch handling is broken). Argh, not again. libdebian-installer 0.22 went into the archive today with some fixes for subarch handling, in particular #243692. Was that version on your image? -- see shy jo signature.asc Description: Digital signature
Re: release status update
On Mon, Mar 08, 2004 at 10:29:28PM -0500, Joey Hess wrote: 3. at least one successful installation report per boot method I have not had time to track installation reports. Could someone take a pass through the list of recent reports and post a summary to the list please? My vague impressions: - i386 is generally working ok - powerpc is working ok the subarches that worked last time - once again we are almost out of time on including any more powerpc subarches No, this is not quite exact. chrp, chrp-rs6k and prep will be working this time around, i think everything is there, but i need to fiddle the build system for it still to build the initrd images, but i do this already by hand, and everything is in place for it. This would use the powerpc netboot initrd. the cdrom initrd is too big though, i will not have time to fix the kernel for this, so maybe building a new specialized initrd image with all the apple stuff removed would do, not sure though. The powerpc-small floppy target will probably not be on time. Jeremie is working on the miboot upload, and once that reaches the archive, the build system can be fixed. power3 and power4/G5 kernels and initrd are not yet done, but i will have a go at them this week, so hopefully that will be ok. Kernels for G5 powerpmac have not been tested, despite my call for volunteer, but if they don't work out, 2.6 kernels will be needed. It needs linux-kernel-di support for the modules, which i hear is problematic, and some volunteer to test them. Here is a short list of things that need to be done still before release: And an update, as if you can't read the top of TODO yourselves. :-) - partman - testing! - clean up lvmcfg UI issue - probably go for i386, powerpc; other arches unknown I tested it on powerpc, altough couldn't do much tests. My disk has vital information i cannot afford to loose. More to this in a separate mail. Friendly, Sven Luther -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: release status update
On Mon, Mar 08, 2004 at 10:29:28PM -0500, Joey Hess wrote: [...] The string freeze has not quite started, since a few strings were changed today. Hopefully we've seen the last of that I also just rewrote most of the strings for the cdebconf text frontend. They should certainly be reviewed by a native English speaker though, one only has to commit changes in cdebconf-text-udeb.templates as usual, and I will patch source files accordingly. Denis -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: release status update
Sven Luther wrote: chrp, chrp-rs6k and prep will be working this time around, i think everything is there, but i need to fiddle the build system for it still to build the initrd images, but i do this already by hand, and everything is in place for it. This would use the powerpc netboot initrd. the cdrom initrd is too big though, i will not have time to fix the kernel for this, so maybe building a new specialized initrd image with all the apple stuff removed would do, not sure though. The powerpc-small floppy target will probably not be on time. Jeremie is working on the miboot upload, and once that reaches the archive, the build system can be fixed. power3 and power4/G5 kernels and initrd are not yet done, but i will have a go at them this week, so hopefully that will be ok. Kernels for G5 powerpmac have not been tested, despite my call for volunteer, but if they don't work out, 2.6 kernels will be needed. It needs linux-kernel-di support for the modules, which i hear is problematic, and some volunteer to test them. Unless I see patches in the archive in the next day or two for the above, I can't imagine you'll make it. -- see shy jo signature.asc Description: Digital signature
Re: release status update (grub notes)
- lowmem package must enter archive - Check archive for udebs that need to be built on various arches; make sure everything is up-to-date. Also, make sure that eg, appropriate kernel debs have entered the archive on each architecture. one thing I noticed with a recent daily build (March 3 I believe is that the GRUB installer does NOT seem to offer the option of putting in other bootable entries, much like the lilo installer in woody would offer the user - Write a release announcement. -- see shy jo -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]