Message ID | cover.1529501006.git.joseph.salisbury@canonical.com |
---|---|
Headers | show |
Series | s390/uprobes: implement arch_uretprobe_is_alive() | expand |
On 06/22/18 18:31, Joseph Salisbury wrote: > BugLink: http://bugs.launchpad.net/bugs/1777840 > > == SRU Justification == > This commit fixes SIGSEGVs with setjmp/longjmp. > > Symptom: > Setting a user space return probe on a function that does not > return because of a longjmp call may cause SIGSEGV or SIGILL. > > Problem: > Userspace can do whatever it wants with its stack. Especially > stack modifications performed with longjmp cannot be handled 100% > correctly within the uprobes kernel part, which may lead to user > space observed crashed. > > Solution: > Implement arch_uretprobe_is_alive() which improves the situation > a lot and works for many situations. > > This commit has also been cc'd to upstream stable, but it has not landed > in Bionic as of yet. > > == Fix == > 783c3b53b950 ("s390/uprobes: implement arch_uretprobe_is_alive()") > > == Regression Potential == > Low. Limited to s390. The commit has also been cc'd to upstream > stable, so it has had additional upstream review. > > == Test Case == > A test kernel was built with this patch and tested by the original bug reporter. > The bug reporter states the test kernel resolved the bug. > > Heiko Carstens (1): > s390/uprobes: implement arch_uretprobe_is_alive() > > arch/s390/kernel/uprobes.c | 9 +++++++++ > 1 file changed, 9 insertions(+) > Hi Joseph, The requested fix has already been applied as part of "Bionic update: upstream stable patchset 2018-06-06" (LP: #1775483), which has already been released so the bug report can be marked as fixed. Thanks, Kleber