Bug#586554: update-initramfs fails to generate initrd.img-2.6.32-5-686 in Debian testing when upgrading from 0.96 to 0.97
Alesh Slovak wrote (Mon, 5 Jul 2010):
On 07/05/2010 09:22 AM, Brendon Higgins wrote:
The test -e command is protected by an , so a failure will not
trigger an early exit and all files will be correctly removed.
I guess you're right. But I tried the script with exit 0 added at the
last line and it failed anyway. I added echo blah statements to find
where it was aborting, and it seemed to be at that test statement. When
I changed it to an if ... fi clause, the problem ceased.
That's odd, our testing went just fine.
Something really odd seems to be going on. It works without errors when I try
this:
awk '{print $2}' $STATE_DIR/clean-files \
| while read file; do
echo ha
test -e ${DESTDIR}$file rm -f ${DESTDIR}$file
echo ho
done
However, if I comment out the echo ho line, it fails. It's like the result
of the test on that last line is erroneously propagating as the result of the
loop, or something.
Can you tell me what shell you are using?
/bin/sh links to dash. I just updated to the latest in unstable - problem
remains. Users see bash.
I also noticed that both files listed in clean-files actually exist, both
before
and after I run update-initramfs. But the rm shouldn't return an error, and
even if it does, adding an echo command after it shouldn't fix it. At least, I
don't see why.
Really weird. :-S
Peace,
Brendon
--
To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/201007052019.36084.blhigg...@gmail.com
Bug#586554: update-initramfs fails to generate initrd.img-2.6.32-5-686 in Debian testing when upgrading from 0.96 to 0.97
On 07/05/2010 07:19 PM, Brendon Higgins wrote:
Something really odd seems to be going on. It works without errors when I try
this:
awk '{print $2}' $STATE_DIR/clean-files \
| while read file; do
echo ha
test -e ${DESTDIR}$file rm -f ${DESTDIR}$file
echo ho
done
However, if I comment out the echo ho line, it fails. It's like the result
of the test on that last line is erroneously propagating as the result of the
loop, or something.
You're right, but it is not erroneous. According to the bash manual page it is
correct behaviour.
The documentation for while states that the exit status of the while and
until commands is the exit status of the last do list command executed, or zero
if none was executed.
With regards to the operator, the manual states that the return status of
AND and OR lists is the exit status of the last command executed in the list.
This together means that the while loop was inheriting the failed status of
the test command and causing the script to exit with a failed status.
The documentation for if states that the exit status is the exit status of
the last command executed, or zero if no condition tested true.
That explains why using an if statement makes the script work as expected.
I couldn't find anything in the dash manual page about all this but it seems to
be working in the same way.
This behaviour appears to be a relatively recent change.
I also noticed that both files listed in clean-files actually exist, both before
and after I run update-initramfs. But the rm shouldn't return an error, and
even if it does, adding an echo command after it shouldn't fix it. At least, I
don't see why.
The files in clean-files exist on your file system, but the script checks for
and erases these files from the initrd file system (i.e. DESTDIR), and not from
your regular file system.
So I think we've finally cleared this up. Thanks for all your help Brendon. It's
really appreciated.
We'll be releasing an updated iscan-data package as soon as we can. I'll reply
here with the details when we do.
Happy scanning,
--
Alesh SlovakLinux Team -- AVASYS Corporation
alesh.slo...@avasys.jp http://avasys.jp
--
To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Archive: http://lists.debian.org/4c328da0.9010...@avasys.jp
Bug#586554: update-initramfs fails to generate initrd.img-2.6.32-5-686 in Debian testing when upgrading from 0.96 to 0.97
On 07/03/2010 10:24 AM, Brendon Higgins wrote: Hi Alesh, Alesh Slovak wrote (Thursday 01 July 2010): If the very last file that iscan's hook script checks for in DESTDIR doesn't exist, the `test -e` call fails and the script ends with a failed status even though nothing really went wrong. A simple `exit 0` on the last line fixes this. It's actually not that simple. Running the iscan hook script with set -e means that the script will abort with an error status as soon as the test -e fails for any file in the clean-files list. Adding exit 0 at the end of the script won't fix that, because the script will never reach that point anyway. I had thought the same at first, but according to the bash man page, the `-e` option exits immediately if any *untested* command fails. A similar statement exists in the dash man page as well. The test -e command is protected by an , so a failure will not trigger an early exit and all files will be correctly removed. Happy scanning, -- Alesh SlovakLinux Team -- AVASYS Corporation alesh.slo...@avasys.jp http://avasys.jp -- To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/4c3116ab.3000...@avasys.jp
Bug#586554: update-initramfs fails to generate initrd.img-2.6.32-5-686 in Debian testing when upgrading from 0.96 to 0.97
Alesh Slovak wrote (2010-07-05 09:18): On 07/03/2010 10:24 AM, Brendon Higgins wrote: Hi Alesh, Alesh Slovak wrote (Thursday 01 July 2010): If the very last file that iscan's hook script checks for in DESTDIR doesn't exist, the `test -e` call fails and the script ends with a failed status even though nothing really went wrong. A simple `exit 0` on the last line fixes this. It's actually not that simple. Running the iscan hook script with set -e means that the script will abort with an error status as soon as the test -e fails for any file in the clean-files list. Adding exit 0 at the end of the script won't fix that, because the script will never reach that point anyway. I had thought the same at first, but according to the bash man page, the `-e` option exits immediately if any *untested* command fails. A similar statement exists in the dash man page as well. The test -e command is protected by an , so a failure will not trigger an early exit and all files will be correctly removed. I guess you're right. But I tried the script with exit 0 added at the last line and it failed anyway. I added echo blah statements to find where it was aborting, and it seemed to be at that test statement. When I changed it to an if ... fi clause, the problem ceased. So I don't know why it was doing that. Peace, Brendon -- To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/201007051022.56651.blhigg...@gmail.com
Bug#586554: update-initramfs fails to generate initrd.img-2.6.32-5-686 in Debian testing when upgrading from 0.96 to 0.97
On 07/05/2010 09:22 AM, Brendon Higgins wrote: The test -e command is protected by an , so a failure will not trigger an early exit and all files will be correctly removed. I guess you're right. But I tried the script with exit 0 added at the last line and it failed anyway. I added echo blah statements to find where it was aborting, and it seemed to be at that test statement. When I changed it to an if ... fi clause, the problem ceased. That's odd, our testing went just fine. Can you tell me what shell you are using? Happy scanning, -- Alesh SlovakLinux Team -- AVASYS Corporation alesh.slo...@avasys.jp http://avasys.jp -- To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/4c312ea8.6090...@avasys.jp
Bug#586554: update-initramfs fails to generate initrd.img-2.6.32-5-686 in Debian testing when upgrading from 0.96 to 0.97
Hi Alesh,
Alesh Slovak wrote (Thursday 01 July 2010):
If the very last file that iscan's hook script checks for in DESTDIR
doesn't exist, the `test -e` call fails and the script ends with a failed
status even though nothing really went wrong. A simple `exit 0` on the
last line fixes this.
It's actually not that simple. Running the iscan hook script with set -e
means that the script will abort with an error status as soon as the test -e
fails for any file in the clean-files list. Adding exit 0 at the end of the
script won't fix that, because the script will never reach that point anyway.
Additionally, this means that the script does not remove all the existing files
listed in clean-files that come after the first file that doesn't exist.
Probably the easiest solution is to just turn that troublesome test -e line
into an if ... fi clause, e.g.:
if test -e ${DESTDIR}$file; then rm -f ${DESTDIR}$file; fi
or something similar.
You might have realised this already, but I felt that it's better to be safe
than sorry. :-)
Peace,
Brendon
signature.asc
Description: This is a digitally signed message part.
Bug#586554: update-initramfs fails to generate initrd.img-2.6.32-5-686 in Debian testing when upgrading from 0.96 to 0.97
* Alesh Slovak alesh.slo...@avasys.jp [Thu Jul 01, 2010 at 11:46:03AM +0900]: On 06/30/2010 07:25 PM, Michael Prokop wrote: We are working on new initramfs-tools version which should display the source of the problem then, I'll be investigating on the iscan issue in more detail as well and report back. Thanks to all for your help tracking this down. I've found the problem. It was indeed related to the fact that the new version of initramfs-tools runs the hook scripts under errexit. If the very last file that iscan's hook script checks for in DESTDIR doesn't exist, the `test -e` call fails and the script ends with a failed status even though nothing really went wrong. A simple `exit 0` on the last line fixes this. I'll try to get a fixed version of iscan-data out as soon as I can. Great, thanks for reporting back. I'll make sure users know which script fails so we can track and debug such issues easier. regards, -mika- signature.asc Description: Digital signature
Bug#586554: update-initramfs fails to generate initrd.img-2.6.32-5-686 in Debian testing when upgrading from 0.96 to 0.97
Package: initramfs-tools Version: 0.97 Severity: normal Hi. Same problem occured here during upgrade from 0.96 to 0.97. Now, I am not even sure the machine will boot properly next time as apparently rebuilding the initramfs failed. r...@halucigenia:~# sh -x /var/lib/dpkg/info/initramfs-tools.postinst + set -e + [ ! -e /etc/initramfs-tools/modules ] + [ x != xtriggered ] + dpkg --compare-versions ge 1.14.5ubuntu10~~ + DPKG_MAINTSCRIPT_PACKAGE= update-initramfs -u update-initramfs: Generating /boot/initrd.img-2.6.32-5-686 update-initramfs: failed for /boot/initrd.img-2.6.32-5-686 Eventually I found the notice in a previous post of this thread saying iscan was the culprit. I purged iscan, but the problem remained. I purged iscan-data, too. Now, finally, it was possible to upgrade initramfs- tools successfully. I have emailed Epson support to inform them of the interference of their drivers with the Debian system. Regards Andreas -- Package-specific info: -- initramfs sizes -rw-r--r-- 1 root root 8.0M Jun 25 18:23 /boot/initrd.img-2.6.32-5-686 -- /proc/cmdline BOOT_IMAGE=/boot/vmlinuz-2.6.32-5-686 root=UUID=69fd34e6-38c8-4a8b-86b8-581daf85ec5d ro quiet -- resume RESUME=UUID=e9012356-b8c2-473e-9547-fa87854b56b2 -- /proc/filesystems ext3 fuseblk -- lsmod Module Size Used by vboxnetadp 5118 0 vboxnetflt 12579 0 vboxdrv 126509 2 vboxnetadp,vboxnetflt acpi_cpufreq4943 0 cpufreq_userspace 1480 0 cpufreq_stats 1940 0 cpufreq_conservative 4018 0 cpufreq_powersave602 0 ppdev 4058 0 lp 5570 0 parport22554 2 ppdev,lp sco 5857 2 bridge 32987 0 stp 996 1 bridge bnep7444 2 rfcomm 25167 0 l2cap 21705 6 bnep,rfcomm crc16 1027 1 l2cap bluetooth 36327 6 sco,bnep,rfcomm,l2cap binfmt_misc 4907 1 uinput 4796 1 fuse 43758 1 loop9757 0 snd_hda_codec_intelhdmi 9027 1 snd_hda_codec_via 19285 1 snd_hda_intel 16703 4 snd_hda_codec 46002 3 snd_hda_codec_intelhdmi,snd_hda_codec_via,snd_hda_intel snd_hwdep 4054 1 snd_hda_codec snd_pcm_oss28671 0 snd_mixer_oss 10461 2 snd_pcm_oss snd_pcm47214 3 snd_hda_intel,snd_hda_codec,snd_pcm_oss snd_seq_midi3576 0 snd_rawmidi12505 1 snd_seq_midi arc4 974 2 snd_seq_midi_event 3684 1 snd_seq_midi ecb 1405 2 snd_seq35463 3 snd_seq_midi,snd_seq_midi_event ath9k 238333 0 mac80211 123258 1 ath9k ath 6014 1 ath9k snd_timer 12258 2 snd_pcm,snd_seq snd_seq_device 3673 3 snd_seq_midi,snd_rawmidi,snd_seq uvcvideo 45226 0 cfg80211 87601 3 ath9k,mac80211,ath i915 219952 2 drm_kms_helper 18305 1 i915 snd34363 18 snd_hda_codec_intelhdmi,snd_hda_intel,snd_hda_codec,snd_hwdep,snd_pcm_oss,snd_mixer_oss,snd_pcm,snd_rawmidi,snd_seq,snd_timer,snd_seq_device soundcore 3450 2 snd rfkill 10264 3 bluetooth,cfg80211 videodev 25545 1 uvcvideo drm 112020 3 i915,drm_kms_helper v4l1_compat10250 2 uvcvideo,videodev snd_page_alloc 5045 2 snd_hda_intel,snd_pcm i2c_algo_bit3497 1 i915 pcspkr 1207 0 asus_laptop11090 0 serio_raw 2916 0 led_class 1757 2 ath9k,asus_laptop i2c_core 12696 5 i915,drm_kms_helper,videodev,drm,i2c_algo_bit tpm_tis 5496 0 video 14605 1 i915 psmouse44653 0 ac 1640 0 battery 3782 0 processor 26599 3 acpi_cpufreq button 3598 1 i915 tpm 8137 1 tpm_tis tpm_bios3569 1 tpm output 1204 1 video evdev 5609 24 ext3 94204 2 jbd32169 1 ext3 mbcache 3762 1 ext3 sg 15968 0 sr_mod 10770 0 cdrom 26487 1 sr_mod sd_mod 25869 4 crc_t10dif 1012 1 sd_mod usbhid 27980 0 hid50629 1 usbhid uhci_hcd 16057 0 ahci 27246 3 thermal 9206 0 libata115665 1 ahci ehci_hcd 27763 0 scsi_mod 101401 4 sg,sr_mod,sd_mod,libata thermal_sys 9378 3 video,processor,thermal atl1e 23144 0 usbcore
Bug#586554: update-initramfs fails to generate initrd.img-2.6.32-5-686 in Debian testing when upgrading from 0.96 to 0.97
* Andreas Neudecker zap...@gmx.net [Wed Jun 30, 2010 at 12:05:37PM +0200]: Same problem occured here during upgrade from 0.96 to 0.97. Now, I am not even sure the machine will boot properly next time as apparently rebuilding the initramfs failed. [...] Eventually I found the notice in a previous post of this thread saying iscan was the culprit. I purged iscan, but the problem remained. I purged iscan-data, too. Now, finally, it was possible to upgrade initramfs- tools successfully. We are working on new initramfs-tools version which should display the source of the problem then, I'll be investigating on the iscan issue in more detail as well and report back. regards, -mika- signature.asc Description: Digital signature
Bug#586554: update-initramfs fails to generate initrd.img-2.6.32-5-686 in Debian testing when upgrading from 0.96 to 0.97
On 06/30/2010 07:25 PM, Michael Prokop wrote: * Andreas Neudeckerzap...@gmx.net [Wed Jun 30, 2010 at 12:05:37PM +0200]: Same problem occured here during upgrade from 0.96 to 0.97. Now, I am not even sure the machine will boot properly next time as apparently rebuilding the initramfs failed. Another iscan maintainer here. I am currently looking into this issue. I purged iscan, but the problem remained. I purged iscan-data, too. Now, finally, it was possible to upgrade initramfs- tools successfully. Starting with iscan 2.25 we've moved bits of the iscan package to a new package, iscan-data. This was done for various reasons, mostly to do with ease of maintenance. The initramfs hook was moved to the new iscan-data package. Since iscan depends on iscan-data, both will need to be uninstalled in order to be able to successfully upgrade mkinitramfs-tools. People with iscan 2.24 or older installed, which does not use iscan-data will just need to uninstall iscan. Happy scanning, -- Alesh SlovakLinux Team -- AVASYS Corporation alesh.slo...@avasys.jp http://avasys.jp -- To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/4c2bddc7.1050...@avasys.jp
Bug#586554: update-initramfs fails to generate initrd.img-2.6.32-5-686 in Debian testing when upgrading from 0.96 to 0.97
On 06/30/2010 07:25 PM, Michael Prokop wrote: We are working on new initramfs-tools version which should display the source of the problem then, I'll be investigating on the iscan issue in more detail as well and report back. Thanks to all for your help tracking this down. I've found the problem. It was indeed related to the fact that the new version of initramfs-tools runs the hook scripts under errexit. If the very last file that iscan's hook script checks for in DESTDIR doesn't exist, the `test -e` call fails and the script ends with a failed status even though nothing really went wrong. A simple `exit 0` on the last line fixes this. I'll try to get a fixed version of iscan-data out as soon as I can. Thanks again. Happy scanning, -- Alesh SlovakLinux Team -- AVASYS Corporation alesh.slo...@avasys.jp http://avasys.jp -- To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/4c2c016b.5020...@avasys.jp
