| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|
|
|
|
|
|
| |
This commit replaces all `ioutil.TempDir` with `t.TempDir` in tests.
The directory created by `t.TempDir` is automatically removed when the
test and all its subtests complete.
Reference: https://pkg.go.dev/testing#T.TempDir
Signed-off-by: Eng Zer Jun <engzerjun@gmail.com>
|
| | |
|
| |
|
|
|
|
| |
All bisections started failing with:
Compiler lacks asm-goto support.
Use gcc 10 for v5.9+.
|
| | |
|
| | |
|
| |
|
|
|
| |
Recipients is not the most importnat part of the vcs interface,
move it from the very top closer to the bottom.
|
| |
|
|
|
|
| |
Returns true if the current tree contains the specified commit
(the commit is reachable from the current HEAD).
Cntains(commit string) (bool, error)
|
| |
|
|
|
| |
Store config options identified using DebugTracer. Also change bisection
and configuration minimization code to use new DebugTracer.
|
| |
|
|
| |
Use vcs.HEAD if syzkaller or kernel commit is not given
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
git fetch can fail due to force-recreated tags:
> ! [rejected] ext4_for_linus -> ext4_for_linus
> (would clobber existing tag)
And due to something related to updating local branches
(see 'git help fetch' for details).
--force should avoid both issues and it seems that
we always want force for our purposes (rather than
fail update/patch testing).
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fetching a remote branch w/o the corresponding remote with git
disables delta optimization on the server's side and the git
client is forced to download the complete branch even if it
already has got an older version. This leads to several problems as:
* Long fetch time
* Limitless growth of local git repo due to git fetch creating
a new pack file every time
Create a remote before fetching to fix the above issues.
With a remote, only the first fetch will take some time and all the following
ones shall be very fast. Furthermore, git fetch will avoid creating many
pack files in .git/objects/pack.
Signed-off-by: Alexander Egorenkov <Alexander.Egorenkov@ibm.com>
|
| |
|
|
|
|
|
|
|
|
|
| |
* pkg/vcs: remove obsolete test script
Per Dmitry, this should have been removed as part of 8f58e4b
("pkg/bisect: switch to kconfig.Minimize").
* all: convert shebang lines to use /usr/bin/env
* Makefile: fix non-portable use of find(1)
|
| |
|
|
|
|
| |
Kconfig depends on the target arch.
Add target argument for Kconfig parsing.
Resolve $(SRCARCH) properly (previously we always assumed x86_64).
|
| |
|
|
|
|
| |
We currently detect v5.10-rc1 as v5.9 because we ignore -rc tags.
This makes it impossible to enable configs that were already added for v5.10.
Treat v5.10-rc1 as v5.10 already.
|
| |
|
|
|
|
|
|
|
|
|
|
| |
We use strings to identify OS/Arch.
These strings are duplicated throughout the code base massively.
golangci-lint points to possiblity of typos and duplication.
We already had to define these names in pkg/csource
and disable checking for prog package. A future change triggers
such warnings in another package.
Add OS/Arch name consts to sys/targets so that they can be used
to refer to OS/Arch. Use the consts everywhere.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Use the new kconfig.Minimize for config minization
instead of the config-bisect.pl script.
This is mostly just deleting code.
Also update tests:
- minimization is now supposed to test the baseline config
(update "testos" stub accordingly)
- minimization is not supposed to return a config that does not build
(a reasonable config minimization procedure can't arrive to such config),
remove test that tests this
Update #2171
|
| |
|
|
|
|
|
| |
Use pkg/kconfig to make changes to .config.
We now have slightly more civilized way of messing with configs.
Update #2171
|
| |
|
|
|
|
|
|
| |
Minimization must call the predicate
(otherwise how does it know this config triggers the crash?).
This does not matter now, but will matter in future.
Update #2171
|
| |
|
|
|
|
|
|
|
|
| |
The pkg/vcs code assumed that we fully manage the repo
within an autonomous program. In particular it tried
to repair any errors by dropping and re-creating the repo.
This does not work well for command-line tools that work
with a user-provided repo. Add OptPrecious for such uses.
Update #2171
|
| |
|
|
|
|
|
| |
Add ReleaseTag method that returns last release tag
for the given commit.
Update #2171
|
| |
|
|
|
|
|
|
| |
The Freshness columns in Instances: table on the dashboard page look outdated, for
these fields are showing when that patch was authored. Where possible, using when
that patch was committed into the tree in question would be more meaningful.
Update #1537
|
| |
|
|
|
|
|
|
|
| |
This config is causing problems to kernel signature calculation. New
initramfs is generated as a part of every build. Due to this init.data
section containing this generated initramfs is differing between
builds causing signture being random number.
Signed-off-by: Jouni Hogander <jouni.hoegander@partner.bmw.de>
|
| |
|
|
|
|
| |
For use inside a Docker container e.g.
Signed-off-by: Alexander Egorenkov <Alexander.Egorenkov@ibm.com>
|
| |
|
|
|
|
|
| |
UBSAN is broken in multiple ways before v5.3, see:
https://github.com/google/syzkaller/issues/1523#issuecomment-696514105
Update #1523
|
| |
|
|
|
|
| |
Add option to use ccache in kernel builds.
Signed-off-by: Jouni Hogander <jouni.hoegander@partner.bmw.de>
|
| |
|
|
|
|
| |
Create a struct on pkg/vcs to store data of syzkaller email recipients
and update its users. The struct contains default name, email, and a
label to divide user into To and Cc when sending the emails.
|
| |
|
|
|
|
|
| |
Fix capitalization, dots at the end
and two spaces after a period.
Update #1876
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Currently we sandbox all repos b/c we assumed
that all builds are also sandboxes. But this causes
havoc for bisection/patch testing b/c syzkaller build
is not actually sandboxed anywhere. Build creates
root-owned files and then git can't do anything with them
but don't report errors either:
$ git checkout 8eda0b95 && echo OK
error: unable to unlink old 'sys/linux/gen/386.go': Permission denied
error: unable to unlink old 'sys/linux/gen/ppc64le.go': Permission denied
...
HEAD is now at 8eda0b957e5b
OK
We trust own sources and we don't test syzkaller patches,
so don't sandbox syzkaller repos.
|
| |
|
|
|
|
|
|
|
|
|
| |
config-bisect.pl uses bash-isms and can't run on non-linux.
It also silently ignores all errors which made failures
very obscure -- the script happily succeeds in presence
of any errors. So the test failed later reading .config.
Use "set -eu" to not fail silently.
Also trace all config-bisect.pl invocations and output.
good/bad decisions are important and we always log them
in the normal bisection.
|
| |
|
|
|
|
| |
/bin/bash is usually installed on linux'es,
but not on other OSes, e.g. *BSDs.
Use /bin/sh instead of /bin/bash.
|
| |
|
|
|
|
|
|
|
|
|
| |
The test requires new git binary.
It's not necessary installed everywhere.
Skip the test if git binary does not support bisection.
Also reoder functions from more important to less important.
Mark the test as parallel.
Move individual tests into subtests.
Split trace per subtest.
|
| | |
|
| |
|
|
|
|
| |
"git clean -fd" does not remove ignored files,
while can mess state when .gitignore changes across commits.
Use "git clean -fdx" to delete ignored files as well.
|
| |
|
|
|
|
|
| |
Change a test to actually produce a new config as it is supposed
to do during config minimization and check the resulting config.
Add a regression test for the bug where we committed a broken config.
|
| |
|
|
|
| |
That part is not linux-specific, move it to the common code.
The more code we have in the common code, the more code we can test.
|
| | |
|
| |
|
|
|
|
|
|
|
| |
Implement Linux kernel configuration bisection. Use bisected minimalistic
configuration in commit bisection. Utilizes config_bisect.pl script from Linux
kernel tree in bisection.
Modify syz-bisect to read in kernel.baseline_config. This is used as a "good"
configuration when bisection is run.
|
| |
|
|
|
| |
At some point it was enabled, but then somehow got disabled.
Re-enable and fix some regressions.
|
| |
|
|
|
| |
Maybe it will help to fix:
https://groups.google.com/forum/#!topic/syzkaller-bugs/2lgvlHd8t1c
|
| | |
|
| |
|
|
|
|
|
|
| |
pkg/vcs/vcs.go:186:2: var `gitSshRepoRe` should be `gitSSHRepoRe`
gitSshRepoRe = regexp.MustCompile(`^(git|ssh|http|https|ftp|ftps)@[a-zA-Z0-9-_]+(\.[a-zA-Z0-9-_]+)+(:[a-zA-Z0-9-_]+)?(/[a-zA-Z0-9-_./]+)?(/)?$`)
^
Signed-off-by: Paul Chaignon <paul.chaignon@gmail.com>
|
| |
|
|
| |
Signed-off-by: Jukka Kaartinen <jukka.kaartinen@unikie.com>
|
| |
|
|
|
|
|
|
|
| |
Jann pointed out that the frame pointer unwinder fails to unwind double
fault stacks. Switch to using the ORC unwinder instead.
https://www.kernel.org/doc/html/latest/x86/orc-unwinder.html
Suggested-by: Jann Horn <jannh@google.com>
|
| |
|
|
| |
Fixes #1604
|
| |
|
|
|
|
|
|
| |
We disabled it in configs, now also disable during bisection.
This config only adds debug output. It should not be enabled at all,
but it was accidentially enabled on some instances for some periods of time,
and kernel is boot-broken for prolonged ranges of commits with deadlock
which makes bisections take weeks.
|
| |
|
|
|
|
| |
BTF fails lots of builds with:
pahole version v1.9 is too old, need at least v1.13
Failed to generate BTF for vmlinux. Try to disable CONFIG_DEBUG_INFO_BTF.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
If you try to run git-using tests while the GIT_DIR environment variable
(and GIT_WORK_TREE, etc) happens to be set, the tests are going to do fun
and exciting things on a repository that isn't the test repository it tries
to set up.
As it turns out, if you try to run "make test" using git rebase -x, you'll
end up with GIT_DIR set to the syzkaller tree. Hilarity ensues.
Unset GIT_DIR, GIT_WORK_TREE and a few other environment variables when
invoking git - that way it'll default to looking at the working directory
that we have given it, which is what we expect.
Signed-off-by: Andrew Donnellan <ajd@linux.ibm.com>
|
| |
|
|
| |
Update #1441
|
| |
|
|
| |
Update #1441
|