Bug #16634
Build patched VeraCrypt packages for buster
100%
Description
In buster we need custom packages for:
- gnome-shell: Patches released in 3.33.2, buster has 3.30.2
- gtk: Not all patches merged yet, will not make it into Buster
- gjs: depends on gtk
- gvfs: The latest of our patches was released in 1.39.1, Buster has 1.38.1
We don’t need custom packages anymore for:
- glib: VeraCrypt patch was released in 2.57.2, Buster has 2.58.3
- udisks2: last VeraCrypt patch released in 2.8.1, which is the version in Buster
- gnome-disk-utility: VeraCrypt patch was released in 3.29.2 , Buster has 3.30.2. We submitted another commit in the context of
Bug #15952to add a tooltip to the keyfile chooser button, which was only released in 3.31.90. But intrigeri wrote onBug #15952that we don’t need to backport this patch (except if the tests turn out to be fragile, which I assume they are not, because intrigeri didn’t reopen the ticket) - gobject-introspection: This was only required because we rebuilt glib. I thought it would also depend on gtk, but that does not seem to be the case.
Subtasks
History
#1 Updated by segfault 2019-04-05 11:36:45
- Feature Branch set to bugfix/16634-build-patched-veracrypt
#2 Updated by segfault 2019-04-05 13:02:30
- blocks Feature #16209: Core work: Foundations Team added
#3 Updated by segfault 2019-04-12 12:21:09
In buster we need custom packages for:
gnome-shell
: Patches not merged yet, will not make it into Bustergtk
: Not all patches merged yet, will not make it into Bustergjs
: depends ongtk
We don’t need custom packages anymore for:
glib
: VeraCrypt patch was released in 2.57.2, Buster has 2.58.3gvfs
: The latest of our patches was released in 1.38.1, which is the version currently in Busterudisks2
: last VeraCrypt patch released in 2.8.1, which is the version in Bustergnome-disk-utility
: VeraCrypt patch was released in 3.29.2 , Buster has 3.30.2. We submitted another commit in the context ofBug #15952to add a tooltip to the keyfile chooser button, which was only released in 3.31.90. But intrigeri wrote onBug #15952that we don’t need to backport this patch (except if the tests turn out to be fragile, which I assume they are not, because intrigeri didn’t reopen the ticket)gobject-introspection
: This was only required because we rebuiltglib
. I thought it would also depend ongtk
, but that does not seem to be the case.
#4 Updated by intrigeri 2019-04-12 12:53:00
> In buster we need custom packages for: […]
> We don’t need custom packages anymore for: […]
Great! Having to maintain only 3 custom packages during the 4.x cycle seems quite tractable.
#5 Updated by segfault 2019-04-12 17:40:00
I built the packages for gtk and gjs today. I want to wait a bit to see if there is more progress on the gnome-shell MR before I backport that. And once I did that I will test the three packages together (gnome-shell depends on gjs and gtk) and then upload them.
#6 Updated by segfault 2019-04-12 17:41:18
- Status changed from Confirmed to In Progress
Applied in changeset commit:tails|461c1397ec4ecb873e5790b5c3c1b30244ba5e3e.
#7 Updated by segfault 2019-06-01 23:29:41
Our gnome-shell patches were merged. I will backport the patches to the version in Buster and build and release them. I hope that I won’t have to do that often before the Tails 4.0 release, I’m a bit worried about that because there have already been 3 migrations of gnome-shell to Buster since the full freeze (see https://tracker.debian.org/pkg/gnome-shell).
#8 Updated by segfault 2019-06-06 07:50:14
I tested the packages by installing them in a feature/buster VM. The gnome-shell unlock dialog doesn’t open. Still have to figure out why.
#9 Updated by segfault 2019-06-06 08:25:17
- Description updated
Above I wrote that we wouldn’t have to patch gvfs anymore, but I was wrong, our latest patch was only released in 1.39.1 and Buster has 1.38.1. Updated the description accordingly.
#10 Updated by segfault 2019-06-06 09:54:56
I’m not sure why, but with the patched gvfs installed, the gnome-shell dialog opens. I fixed another issue with the gnome-shell patches and now everything seems to work. Releasing the packages now.
#11 Updated by segfault 2019-06-06 10:12:21
I released and uploaded gnome-shell 3.30.2-9.0tails1 and pushed a new branch tails/buster
to the existing gnome-shell
repository on git.tails.boum.org.
@intrigeri: do you prefer it if I continue using my personal packaging repositories on gitlab.com for the other packages or can I create repositories for them on git.tails.boum.org as well?
#12 Updated by intrigeri 2019-06-07 09:53:35
> intrigeri: do you prefer it if I continue using my personal packaging repositories on gitlab.com for the other packages or can I create repositories for them on git.tails.boum.org as well?
Now that it’s clear that we’ll have to maintain these patched packages for 2 more years (Tails 4.x), it would be much better to have the packaging live in some place where all Tails commiters have access, instead of just one person. git.tails.b.o is OK (that’s where most of our code currently lives) but given our plans to migrate to GitLab, I’d rather create these new repos directly under https://salsa.debian.org/tails-team/ (optimistically assuming we’ll settle on Salsa is our GitLab instance of choice) instead of adding more and more repos to the list of those we’ll want to migrate later. Happy to do whichever you prefer, just tell me :)
#13 Updated by segfault 2019-06-09 07:46:13
intrigeri wrote:
> I’d rather create these new repos directly under https://salsa.debian.org/tails-team/ (optimistically assuming we’ll settle on Salsa is our GitLab instance of choice) instead of adding more and more repos to the list of those we’ll want to migrate later. Happy to do whichever you prefer, just tell me :)
Agreed, let’s use GitLab. Thanks!
#14 Updated by intrigeri 2019-06-09 16:23:23
> Agreed, let’s use GitLab. Thanks!
If you are not allowed to created the repos yourself, please give me a list of package names and I’ll create them.
#15 Updated by segfault 2019-06-12 20:05:54
intrigeri wrote:
> If you are not allowed to created the repos yourself, please give me a list of package names and I’ll create them.
That’s gtk, gvfs, and gjs. I created a repo for gtk but I’m not allowed to push to it, it’s rejected because there is no master branch and I’m not allowed to create one. Also, I would prefer to fork the existing Debian packaging repos via Gitlab, but I’m not allowed to fork them into the tails-team namespace.
#16 Updated by segfault 2019-06-13 12:43:53
Rebuilt and released gvfs because debian/1.38.1-5 was released (previous build was based on debian/1.38.1-3).
I now forked the Debian packaging repos for gtk3, gjs, and gvfs into my own namespace (segfault-guest) and pushed the tags there. @intrigeri: You should be able to just fork my repos into the tails-team namespace.
#17 Updated by segfault 2019-06-13 19:20:14
Uploading the gtk3 .changes file fails with “Cannot put file ‘libgtk-3-0-udeb_3.24.5-1.0tails1_amd64.udeb’ of ‘gtk+3.0_3.24.5-1.0tails1_amd64.changes’ into component ‘main’, as it is not listed in UDebComponents of ‘bugfix-16634-build-patched-veracrypt’”. intrigeri, I assume that you worked around this issue when you uploaded the old gtk3 package, which also included a udeb. I don't see a UDebComponents field anywhere in
conf/distributions@, so I guess that’s not the workaround you chose.
#18 Updated by intrigeri 2019-06-14 08:39:45
> Uploading the gtk3 .changes file fails with “Cannot put file ‘libgtk-3-0-udeb_3.24.5-1.0tails1_amd64.udeb’ of ‘gtk+3.0_3.24.5-1.0tails1_amd64.changes’ into component ‘main’, as it is not listed in UDebComponents of ‘bugfix-16634-build-patched-veracrypt’”. intrigeri, I assume that you worked around this issue when you uploaded the old gtk3 package, which also included a udeb. I don’t see a UDebComponents field anywhere in conf/distributions
, so I guess that’s not the workaround you chose.
Indeed, you need to remove the udebs from the .changes
file before uploading.
#19 Updated by intrigeri 2019-06-14 09:16:09
> I created a repo for gtk but I’m not allowed to push to it, it’s rejected because there is no master branch and I’m not allowed to create one.
OK, deleted this repo (and the empty gjs one).
> Also, I would prefer to fork the existing Debian packaging repos via Gitlab
Same.
#20 Updated by intrigeri 2019-06-14 09:17:48
> I now forked the Debian packaging repos for gtk3, gjs, and gvfs into my own namespace (segfault-guest) and pushed the tags there. intrigeri: You should be able to just fork my repos into the tails-team namespace.
I’d rather fork the GNOME team’s repos so the forks graph is cleaner (and then you can probably drop your own forks). I’ll do that and push your tags/code to our fork.
#21 Updated by intrigeri 2019-06-14 09:37:18
intrigeri wrote:
> I’d rather fork the GNOME team’s repos so the forks graph is cleaner (and then you can probably drop your own forks). I’ll do that and push your tags/code to our fork.
Done! You should have push access (at least to existing branches) on these new repos.
#22 Updated by segfault 2019-06-14 15:14:58
- Status changed from In Progress to Needs Validation
- Assignee deleted (
segfault)
intrigeri wrote:
> Done! You should have push access (at least to existing branches) on these new repos.
Thanks! I deleted my personal repos.
I uploaded all packages now. FTR, to upload GTK I had to delete the udeb entries and resign the .changes file.
I built the feature branch and the VeraCrypt feature seems to work.
#23 Updated by intrigeri 2019-06-15 09:34:41
@segfault, can you please merge current feature/buster into the topic branch, so that Jenkins can build it? (Currently it fails when it tries to merge devel into your branch, and that merge conflict was resolved in feature/buster already.)
#24 Updated by segfault 2019-06-16 10:04:20
intrigeri wrote:
> @segfault, can you please merge current feature/buster into the topic branch, so that Jenkins can build it? (Currently it fails when it tries to merge devel into your branch, and that merge conflict was resolved in feature/buster already.)
Done
#25 Updated by intrigeri 2019-06-17 07:31:29
- Assignee set to intrigeri
#26 Updated by intrigeri 2019-06-17 07:42:28
- Status changed from Needs Validation to In Progress
I see a scenario failing (https://jenkins.tails.boum.org/view/Tails_ISO/job/test_Tails_ISO_bugfix-16634-build-patched-veracrypt/1/cucumberTestReport/using-veracrypt-encrypted-volumes/use-unlock-veracrypt-volumes-to-unlock-a-hidden-veracrypt-file-container/). The screenshot looks like everything is working well. The image Sikuli fails to find is fuzzy in other scenarios (that pass) anyway so I’ll update it and we’ll see.
#27 Updated by intrigeri 2019-06-17 08:31:30
- Status changed from In Progress to Resolved
- % Done changed from 0 to 100
Applied in changeset commit:tails|fb169ca1c3f6abe063a1272053eadca79945e76c.