Debian Patches
Status for snapd/2.76.3-2
| Patch | Description | Author | Forwarded | Bugs | Origin | Last update |
|---|---|---|---|---|---|---|
| 0003-cmd-snap-seccomp-skip-tests-that-use-m32.patch | [PATCH 3/9] cmd/snap-seccomp: skip tests that use -m32 Apparently Debian's amd64 compiler somehow cannot compile -m32 mode binaries. The compilation error is: multipass@debian-10:~/packaging/snapd/cmd/snap-seccomp$ go test cannot build multi-lib syscall runner: exit status 1 In file included from /usr/include/errno.h:25, from /tmp/check-3806730340354206876/1/seccomp_syscall_runner.c:3: /usr/include/features.h:424:12: fatal error: sys/cdefs.h: No such file or directory # include <sys/cdefs.h> ^~~~~~~~~~~~~ compilation terminated. OK: 2 passed, 11 skipped I was unable to resolve this issue, let's disable this test until we can get to the bottom of it. |
Zygmunt Krynicki <me@zygoon.pl> | no | 2019-01-17 | ||
| 0004-cmd-snap-skip-tests-depending-on-text-wrapping.patch | [PATCH 4/9] cmd/snap: skip tests depending on text wrapping Upstream snapd contains tests that check the output of various commands along with the --help command-line argument. The output is wrapped to match terminal width and for readability. The algorithm for wrapping has apparently changed across versions of github.com/jessevdk/go-flags. Since this test is not critical for anything it can be disabled to let the package build. |
Zygmunt Krynicki <me@zygoon.pl> | no | 2019-01-17 | ||
| 0005-tests-adjust-gendeveloper1-to-older-github.com-jesse.patch | tests: adjust gendeveloper1 to older github.com/jessevdk/go-flags | Zygmunt Krynicki <me@zygoon.pl> | no | upstream | 2025-11-21 | |
| 0010-man-page-sections.patch | man-page-sections | Michael Hudson-Doyle <mwhudson@debian.org> | no | 2026-06-30 | ||
| 0011-overlord-install-fix-flaky-TestEncryptionAvailabili.patch | overlord/install: fix flaky TestEncryptionAvailabilityCheck timeout mock The ErrorSecbootTimeout mocks for secboot.PreinstallCheck and secboot.PreinstallCheckAction simulated a slow check by racing a ```go select { case <-ctx.Done(): return nil, nil, ctx.Err() case <-time.After(20 * time.Millisecond): } ``` against a context whose timeout is mocked down to 1ms. Under normal load ctx.Done() always wins by a wide margin, but on a slow or heavily loaded machine, scheduling delays for the whole process can exceed the 20ms gap, letting both timers become ready by the time select is evaluated. Go picks pseudo-randomly among ready cases in that situation, so the mock occasionally took the "success" branch instead of returning ctx.Err(), making encryptionAvailabilityCheck return a non-nil check context where the test expected nil: ```go obtained *secboot.PreinstallCheckContext = &secboot.PreinstallCheckContext{} expected *secboot.PreinstallCheckContext = (*secboot.PreinstallCheckContext)(nil) ``` This was observed as an intermittent failure on a Debian s390x buildd. Fix it by waiting deterministically on ctx.Done() instead of racing it against an unrelated timer: the mock only needs to simulate "the check took too long", not actually perform concurrent work that can be cancelled. |
Zygmunt Krynicki <zygmunt.krynicki@canonical.com> | yes | vendor | 2026-08-21 |
All known versions for source package 'snapd'
- 2.76.3-2 (sid)
- 2.76-1 (forky)
- 2.68.3-3+deb13u1 (trixie-proposed-updates, trixie, trixie-security)
- 2.57.6-1+deb12u1 (bookworm-security, bookworm)
