This fixes the others/crit test to work again and extends it to make
sure all possible input and output options are correctly handled by
crit.
Signed-off-by: Adrian Reber <areber@redhat.com>
CI sometimes errors out encoding/decoding extra pipe data.
This should fix extra pipe data for Python 3 and still keep it working
on Python 2.
Signed-off-by: Adrian Reber <areber@redhat.com>
This changes stdin to be opened as binary if the input is not a tty.
This changes stdout to be opened as binary if encoding or if the output
is not a tty.
Signed-off-by: Adrian Reber <areber@redhat.com>
In image dump directory, there are 2 parent symlink error cases:
- Parent symlink does not exist
- Parent symlink exists but points to invalid target
At the moment, 2 cases are handled exactly the same (do full dump). However, while
the first case happen when parent path is not provided, the second one is likely
user's mistake when provides invalid parent path.
So we throw an error in the latter case instead of performing the full dump.
Signed-off-by: Bui Quang Minh <minhquangbui99@gmail.com>
Closes: #1408
CRIU_REQ_TYPE__FEATURE_CHECK was failing, this was caused by two
things in handle_feature_check():
1. setup_opts_from_req() was used and it could be NULL
(kerndat_init() is enough for feature checking)
2. resp.success was always set to false
Signed-off-by: Zeyad Yasser <zeyady98@gmail.com>
The recent fix to make Jenkins run crit-recode again broke
Python 2 support (because Python 2 based CI was not running).
This should fix the Python 2 based test run.
Signed-off-by: Adrian Reber <areber@redhat.com>
Previously we only skipped replacing sid for shell job if root_item was
session leader, but there is other case where root_item sid is the same
as current_sid we can safely skip replacing for this case. Same applies
to gid-s.
Now after we have pid collision check we not only "can" but should skip
pid collision checks for the latter case. It is quite obvious that
there are tasks in tree with sid==current_sid if current_sid==old_sid.
Fixes: #1400
Fixes: 77968d43c ("pstree: check for pid collision before switching to
new sid/gid")
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
With the switch to Python3 and binary output it is not possible to use
code like: 'f.write('\0' * (rounded - size))'. Switching to binary
helps.
Signed-off-by: Adrian Reber <areber@redhat.com>
fromstring() and tostring() are deprecated since Python 3.2 and have
been removed in 3.9. Both functions were just aliases and this patch
changes images.py to directly call fromybytes() and tobytes().
Signed-off-by: Adrian Reber <areber@redhat.com>
The criu-ns script creates a new PID namespace where criu is the "init"
process. When using the --restore-detached option with criu-ns, users
expect criu-ns to exit without killing the restored process tree.
Thus, criu-ns should not pass the --restore-detached to criu to prevent
it from terminating, and it should exit instead of waiting for criu's
exit status.
Resolves#1278
Suggested-by: Andrei Vagin <avagin@gmail.com>
Signed-off-by: Radostin Stoyanov <rstoyanov@fedoraproject.org>
This change extracts some of the duplicated code from
set_pidns() and set_mntns() functions.
Signed-off-by: Radostin Stoyanov <rstoyanov@fedoraproject.org>
This test should be run with at least 1 pre-dump to trigger the problem as mentioned in commit 4d9bf608b5.
Signed-off-by: Zeyad Yasser <zeyady98@gmail.com>
ccache was set up in Travis to speed up compilation by re-using the
.ccache directory from previous CI runs. As we are no longer using
Travis we can remove all CI related ccache setup.
Signed-off-by: Adrian Reber <areber@redhat.com>
We were running crit-recode in CI only on the output of
zdtm/static/env00.
This adds zdtm/transition/fork and zdtm/static/ghost_holes00
to run through crit-recode as the image files from those test
triggered errors in Jenkins we did not see in CI.
Signed-off-by: Adrian Reber <areber@redhat.com>
Although we are running crit-recode.py also in all CI runs we never seen
following error except in Jenkins:
Traceback (most recent call last):
File "/usr/lib/python3.8/base64.py", line 510, in _input_type_check
m = memoryview(s)
TypeError: memoryview: a bytes-like object is required, not 'str'
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "./test/crit-recode.py", line 25, in recode_and_check
r_img = pycriu.images.dumps(pb)
File "/var/lib/jenkins/workspace/Q/test/pycriu/images/images.py", line 635, in dumps
dump(img, f)
File "/var/lib/jenkins/workspace/Q/test/pycriu/images/images.py", line 626, in dump
handler.dump(img['entries'], f)
File "/var/lib/jenkins/workspace/Q/test/pycriu/images/images.py", line 289, in dump
f.write(base64.decodebytes(item['extra']))
File "/usr/lib/python3.8/base64.py", line 545, in decodebytes
_input_type_check(s)
File "/usr/lib/python3.8/base64.py", line 513, in _input_type_check
raise TypeError(msg) from err
TypeError: expected bytes-like object, not str
This commit fixes this by encoding the string to bytes.
Signed-off-by: Adrian Reber <areber@redhat.com>
This removes extending LDFLAGS with '-Wl,-z,now'. This was added as
workaround but never really worked. It is correctly fixed with
pull request #1379
Signed-off-by: Adrian Reber <areber@redhat.com>
python3 fails to encode image with the following:
> [dima@Mindolluin criu]$ ./crit/crit encode -i tmp -o tmp.1
> Traceback (most recent call last):
> File "/home/dima/src/criu/./crit/crit", line 6, in <module>
> cli.main()
> File "/home/dima/src/criu/crit/pycriu/cli.py", line 410, in main
> opts["func"](opts)
> File "/home/dima/src/criu/crit/pycriu/cli.py", line 50, in encode
> pycriu.images.dump(img, outf(opts))
> File "/home/dima/src/criu/crit/pycriu/images/images.py", line 617, in dump
> f.write(struct.pack('i', magic.by_name['IMG_COMMON']))
> TypeError: write() argument must be str, not bytes
Opening the output file as binary seems to help.
Signed-off-by: Dmitry Safonov <dima@arista.com>
By default docker writes logs in a run-time directory unique for each
container. To be able to read this file, we can specify the path in
CRIU's configuration file for runc.
Signed-off-by: Radostin Stoyanov <rstoyanov@fedoraproject.org>
Without this check we can hit the BUG in lookup_create_item just a few
steps later (if one thread in images has same pid with new sid/gid). And
also this check saves us from different sorts of unexpected errors on
restore (if one non-thread task in images has same pid/sid/gid already).
Fixes: #1332
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Our Jenkins s390x vt test fails with
./vt --pidfile=vt.pid --outfile=vt.out --filename=vt.test
make: *** [Makefile:432: vt.pid] Error 1
Test zdtm/static/vt FAIL at ['make', '--no-print-directory', '-C', 'zdtm/static', 'vt.pid']
Test output: ================================
08:08:15.556: 54: ERR: vt.c:38: Open virtual terminal vt.test failed (errno = 6 (No such device or address))
08:08:15.556: 53: ERR: test.c:316: Test exited unexpectedly with code 1
Because the host has no ttyS0 as used previously. This changes the test
to use 'ttysclp0'. That seems to exist on multiple s390x we checked.
Signed-off-by: Adrian Reber <areber@redhat.com>
This reverts commit c98af78c58.
Now FPU/SSE/MMX/etc can be used inside parasite.
Let's have compiler optimizations back.
Cc: Alexander Mikhalitsyn <alexander.mikhalitsyn@virtuozzo.com>
Signed-off-by: Dmitry Safonov <dima@arista.com>
Arch-dependend way to restore extended registers set.
Use it straight-away to restore per-thread registers.
Signed-off-by: Dmitry Safonov <dima@arista.com>
Extended registers set for task is restored with rt_sigreturn() through
prepared sigframe. For threads it's currently lost.
Preserve it inside thread context to restore on thread curing.
Signed-off-by: Dmitry Safonov <dima@arista.com>
CRIU dumps main thread and sub-threads differently, so there needed
a test to check if fpu is preserved across C/R in sub-threads.
Signed-off-by: Dmitry Safonov <dima@arista.com>
With pseudo-random garbage, the seed is printed with pr_err().
get_task_regs() is called during seizing the task and also for each
thread.
At this moment only for x86.
Signed-off-by: Dmitry Safonov <dima@arista.com>
If snprintf was truncated we should probably know about it instead of
continuing to increase off, as snprintf returns number of characters
which would have been written and not the number which was actually
written.
Normally we check snprintf only for overflow not for error, some modern
compilers print warnings if truncation was not checked.
Probably it was the case why we implemented [1], the truncation happened
and on the next iteration of for loop we've hit negative size for
snprintf and got -1.
Fixes: 90f043dea ("namespaces: handle errors of snprintf") [1]
Signed-off-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
To minimize things done in parasite, PTRACE_GET_THREAD_AREA can be
used to get remote tls. That also removes an additional compat stack
(de)allocation in the parasite (also asm-coded syscall).
In order to use PTRACE_GET_THREAD_AREA, the dumpee should be stopped.
So, let's move this from criu to compel to non-seized state and put tls
into thread info on x86.
Signed-off-by: Dmitry Safonov <dima@arista.com>
It seems the Fedora rawhide /tmp is no longer 1777 but 755.
Change it back to 1777 to make our CI runs successful again.
Signed-off-by: Adrian Reber <areber@redhat.com>
