Bug #7938
Universal USB Installer creates unbootable devices
10%
Description
After installing Tails-1.1.1l using UUI 1.9.5.6, when trying to boot users get that error message:
SYSLINUX 4.07… Avin et al
ifcpu64.32: not a COM32R Image
Answers to ‘Tails does not start’.
1. Tails 1.1.1
2. ISO image verified using Gpg4Win.
3. computer is a ACPI x64 - based PC from IBUYPOWER.
4. See above
5. Tails installed on a 32 GB Sony USB
6. No.
7. Can’t try.
8. Can’t try.
9. Installing Tails 1.1.1 ISO using UUI.
Files
Subtasks
Related issues
Blocks Tails - |
Rejected | 2016-01-22 | |
Blocks Tails - |
Rejected | 2016-09-22 |
History
#1 Updated by sajolida 2014-09-23 09:59:25
I tried to install Tails 1.1.1 using UUI 1.9.5.6 and couldn’t reproduce that. The install worked as expected.
#2 Updated by mercedes508 2014-09-24 01:09:17
- Status changed from New to Confirmed
#3 Updated by mercedes508 2014-09-24 01:12:29
- Status changed from Confirmed to New
#4 Updated by eillip 2014-09-27 07:57:16
- File 7938_workaround.diff added
Hi,
I ran into this bug and it looks related to the feature allowing to choose a 64bit kernel.
I patched the syslinux/syslinux.cfg file (patch attached) to remove the call to ifcpu64.c32 and change the default boot option and Tails booted without problem. This patch remove the 64bits kernel option so it is more a workaround than a real fix. I would be happy to test any other patch to fix this problem.
If I am not mistaken this bug will likely only touch users of 32bits kernels and is totaly unrelated to the Tails installer.
#5 Updated by intrigeri 2014-11-10 13:20:37
> SYSLINUX 4.07
This should not happen. A proper installation of Tails with UUI should install syslinux 6.x. Not sure what’s happening here, as the log looks as if everything went fine. I suspect UUI mixes, in some cases, c32 modules that come from its own distribution (4.x) with those that we ship (6.x)
eillip wrote:
> I ran into this bug
Did you see “SYSLINUX 4.07” too?
> I patched the syslinux/syslinux.cfg file (patch attached) to remove the call to ifcpu64.c32 and change the default boot option and Tails booted without problem. This patch remove the 64bits kernel option so it is more a workaround than a real fix. I would be happy to test any other patch to fix this problem.
Thanks, this is helpful.
Can you please provide the output of ls -lAR /lib/live/mount/medium
on that system, and tell us what version of Tails it is currently running?
> If I am not mistaken this bug will likely only touch users of 32bits kernels
I doubt it: all users need to go through ifcpu64.c32
.
#6 Updated by intrigeri 2014-11-10 13:21:01
- QA Check set to Info Needed
#7 Updated by goupille 2014-12-07 11:47:17
A user reported having the same bug (SYSLINUX 4.07 error) with UUI (Version 1.9.5.8) and Tails 1.2.1, however, I couldn’t reproduce it myself. Eventually, the user has been able to boot on that USB stick, after installing Tails with Tails Installer.
#8 Updated by kosh16 2014-12-28 20:53:24
UUI starts the program syslinux.exe during execution. If the execution of syslinux.exe is blocked by a virussoftware, the syslinux4.07 error occurs. To avoid unbootable USBsticks the sentence ‘Turn off your virusscanner.’ should be added here: https://tails.boum.org/doc/first_steps/installation/manual/windows/index.en.html
I don’t know, if it is part of this ticket but UUI does not PARTITION the USB stick properly. It’s expecting an MBR partitioned stick!!
#9 Updated by intrigeri 2015-01-15 15:48:23
- Assignee set to goupille
I suspect that this bug might happen only when the target device already has syslinux 4.x installed. goupille, may you please try installing another Live system supported by UUI on a UBS stick, and then trying to install Tails with UUI on the same device, without formatting/resetting the device in the meantime?
#10 Updated by BitingBird 2015-04-10 19:27:32
goupille, any news on this?
#11 Updated by goupille 2015-04-12 16:27:35
sorry, I forgot about that. SO yesterday I tested that with the current version of UUI (1.9.5.9).
I formated the key and installed a Puppy 4.3.1 iso on the stick, tested it, it worked. Then, I installed Tails 1.3.2 without checking the “format the device” box, tested it, and it worked.
I tried again with Kali 1.0.6 after formatting the stick, booted on it (it uses Syslinux 4.0.7), installed Tails 1.3.2 without erasing the data, and it booted (using syslinux 6.0.3).
so either this was not the issue or it was corrected in UUI 1.9.5.9.
However, I encounter another issue, at first : when installing any iso on a stick that has a partition table in it (like a Tails device created with Tails installer) UUI install Tails on the first partition (/dev/sdb1) and the computer don’t see it at boot (“no operating system”. To make it work I had to erase the partition table and format the whole volume (/dev/sdb) in FAT32.
#12 Updated by intrigeri 2015-04-25 06:41:16
> yesterday I tested that with the current version of UUI (1.9.5.9).
Thanks :)
> I formated the key
What do you mean exactly? Your comment below about formatting the whole volume in FAT32 makes me wonder.
> I tried again with Kali 1.0.6 after formatting the stick,
Same question here.
We’ll see once goupille clarifies, but I’m starting to wonder if we should perhaps document what users should do exactly wrt. the “format the device” checkbox. What does this checkbox do exactly?
#13 Updated by goupille 2015-04-27 10:02:50
when I said that “I formatted” the stick before installing Puppy and Kali, I used Gnome Disk Utility to erase all the partitions, create a new one on /dev/sdb and format it in FAT32.
from what I remember, during my tests, checking the box seemed to make UUI format in FAT32 (it was quick, so I think without erasing anything) the first partition (I don’t think I would have the “no operating system” error if it formated the whole volume)
#14 Updated by goupille 2015-04-28 07:38:01
I’m doing more testing about UUI today. I did all the “formatting” and partition managment with Gnome Disks Utility, except if it is said otherwise. for the step 1 and 2 I used a PNY 8G, then I used a nearly new Transcend 16 Go USB2 stick (that once worked with Tails installed by Tails Installer) :
1. installing Tails with UUI on a stick previously installed with Tails installer (“format box” unchecked)-> fails, UUI uses only the first partition and ignore the space left for the persistence volume, no error messages from UUI, after that windows can’t recognize the device at all.
2. installing Tails with UUI on a stick previously installed with Tails installer (“format box” checked)-> fails, UUI uses only the first partition and ignore the space left for the persistence volume, an error in the logs “Failed to get parition info”, followed by “GetLastError()=1: Incorrect Function” (“parition” is misspelled in the logs)
3. installing Tails with UUI on a stick formatted with ext4 (no partition) (“format box” checked or not, same result)-> fails with numerous errors from UUI (“the device will no be bootable”, “the iso is not compatible with UUI”, no permission to write on the device…)
4. installing Tails with UUI on a stick formatted with ext4 (with a partition table and one partition) (“format box” checked or not, same result) -> fails, same results, device is write protected.
in the last two cases, Windows can’t format the sticks because of the write permissions.
I tried again 3 and 4 with another stick (Lexar USB 3.0), with the same results.
5. installing Tails with UUI on the Transcend, formatted with fat32 (with a partition table and one partition) (“format box” checked or unchecked, same result) -> fails, absolutly no errors in UUI’s logs, but the BIOS don’t see the device as bootable.
however I tried again with the Lexar and it worked.
I think that there was an error during one of the formatting operations on the Transcend stick because Tails Installer couldn’t reinstall Tails on it because of differences between the boot sector and its backup, and that :
/tmp/tmpFQZ7VJ/syslinux -d syslinux /dev/sdb1
Can’t open /proc/14175/fd/3: Permission denied
Cannot initialize ‘S:’
Bad target s:/ldlinux.sys
/tmp/tmpFQZ7VJ/syslinux: failed to create ldlinux.sys
so, after these tests, I think I can say that UUI can’t handle a stick with a non windows filesystem. otherwise, I don’t know anything more about UUI’s behaviour… I don’t see any advice we could give that could reduce the amount of errors, without entring in the “tech zone”.
#15 Updated by intrigeri 2015-06-18 10:09:19
Subject: Universal USB Installer allows installation scenarios that cannot possibly work
Thanks a lot for the thorough testing session!
… and sorry for the delay :/ It would be awesome if the frontdesk / code contributors interface wasn’t limited to “any developer who follows all Redmine activity” (that is, in practice: me). Food for thought, let’s discuss that at the summit.
> I’m doing more testing about UUI today. I did all the “formatting” and partition managment with Gnome Disks Utility, except if it is said otherwise. for the step 1 and 2 I used a PNY 8G, then I used a nearly new Transcend 16 Go USB2 stick (that once worked with Tails installed by Tails Installer) :
> 1. installing Tails with UUI on a stick previously installed with Tails installer
> (“format box” unchecked)-> fails, UUI uses only the first partition and ignore the
> space left for the persistence volume, no error messages from UUI,
Does “fails” mean that the resulting USB stick doesn’t boot?
Anyway, I didn’t expect this use case to work at all, due to the way Tails Installer partitions the device. But since UUI pretends to do its job, but actually fails, it would be useful to document better what kind of destination devices are supported or not.
> 2. installing Tails with UUI on a stick previously installed with Tails installer
> (“format box” checked)-> fails, UUI uses only the first partition and ignore the
> space left for the persistence volume, an error in the logs “Failed to get parition
> info”, followed by “GetLastError()=1: Incorrect Function” (“parition” is misspelled
> in the logs)
Same here, I don’t think UUI supports GPT devices so no big surprise it doesn’t work. At least it fails in a noisy way that doesn’t let the user believe that the resulting device will work.
> 3. installing Tails with UUI on a stick formatted with ext4 (no partition) (“format
> box” checked or not, same result)-> fails with numerous errors from UUI (“the device
> will no be bootable”, “the iso is not compatible with UUI”, no permission to write on
> the device…)
OK. We don’t support booting Tails from ext4 anyway, but again, if UUI allows one to try that, then we should document that it won’t work.
> 4. installing Tails with UUI on a stick formatted with ext4 (with a partition table
> and one partition) (“format box” checked or not, same result) -> fails, same results,
> device is write protected.
Same as case 3.
> 5. installing Tails with UUI on the Transcend, formatted with fat32 (with a partition
> table and one partition) (“format box” checked or unchecked, same result) -> fails,
> absolutly no errors in UUI’s logs, but the BIOS don’t see the device as bootable.
Was in a MBR or GPT partition table? If GPT, then I’m not overly surprised that UUI fails at it. If MBR, then that’s a problem: this is exactly the use case we have for UUI.
> however I tried again with the Lexar and it worked.
OK…
> I think that there was an error during one of the formatting operations on the
> Transcend stick because Tails Installer couldn’t reinstall Tails on it because of
> differences between the boot sector and its backup, and that :
> /tmp/tmpFQZ7VJ/syslinux -d syslinux /dev/sdb1
> Can’t open /proc/14175/fd/3: Permission denied
> Cannot initialize ‘S:’
> Bad target s:/ldlinux.sys
> /tmp/tmpFQZ7VJ/syslinux: failed to create ldlinux.sys
That sounds like a totally different bug in Tails Installer. If you manage to reproduce it, please file a dedicated bug report. Otherwise, no big deal, I’ve never heard of such a problem, so the pre-conditions to reproduce it are probably rarely met.
=> I believe we should “simply” document that UUI must be used:
- on a MBR partitioned device
and
- selecting either a fresh, empty FAT32 partition as the destination, or checking the “Format” box in UUI.
And this should get rid of all these problems. Could you please confirm that these two use cases (that I’d like to support) actually work?
#16 Updated by goupille 2015-06-20 11:15:28
Hi !
> Does “fails” mean that the resulting USB stick doesn’t boot?
yes
> Was in a MBR or GPT partition table? If GPT, then I’m not overly surprised that UUI fails at it. If MBR, then that’s a problem: this is exactly the use case we have for UUI.
I can’t remember, but since Disk Utility is proposing MBR by default…
> => I believe we should “simply” document that UUI must be used:
>
> on a MBR partitioned device
>
> and
>
> selecting either a fresh, empty FAT32 partition as the destination, or checking the “Format” box in UUI.
>
> And this should get rid of all these problems. Could you please confirm that these two use cases (that I’d like to support) actually work?
I agree with those clarifications, I can try to test these use case, but not before july…
#17 Updated by BitingBird 2016-06-29 06:44:41
- Status changed from New to In Progress
- % Done changed from 0 to 10
#18 Updated by goupille 2016-08-19 12:49:44
- Subject changed from Universal USB Installer (1.9.5.6) creates unbootable devices to Universal USB Installer creates unbootable devices
- Assignee changed from goupille to sajolida
maybe it would be a good idea to run a serie of test to see how UUI is behaving these days, and reject this bug if there are no news ?
#19 Updated by sajolida 2016-08-24 01:48:06
- Target version set to Tails_2.6
#20 Updated by sajolida 2016-09-13 05:05:53
- Assignee changed from sajolida to goupille
- Target version deleted (
Tails_2.6)
I will most likely not be able to do this before November, so goupille, maybe you want to find someone else with a Windows installed.
#21 Updated by sajolida 2016-09-24 06:10:13
- related to
Feature #10984: Test Rufus added
#22 Updated by sajolida 2016-09-24 06:10:42
- related to deleted (
)Feature #10984: Test Rufus
#23 Updated by sajolida 2016-09-24 06:11:33
- Type of work changed from Research to Test
I bet that anybody willing to help goupille on testing this would be more than welcome.
#24 Updated by sajolida 2016-09-24 06:11:45
- blocks
Feature #10984: Test Rufus added
#25 Updated by sajolida 2016-09-25 10:15:28
- blocks
Feature #11825: Consider switching to Rufus instead of UUI for Windows users added
#26 Updated by sajolida 2016-11-16 12:32:15
- Assignee changed from goupille to sajolida
UUI 1.9.6.8 worked fine to install 2.7.
#27 Updated by sajolida 2016-12-05 13:17:53
- Status changed from In Progress to Rejected
- Assignee deleted (
sajolida)
Rejecting this until someone comes up with a way to reproduce it.