In a stack overflow test from the Fuchsia tree, an intentional crash was
being induced that at the point it was reported to Crashpad resulted in
a stack pointer outside of the stack. This caused two problems:
- Crashpad attempted to capture that whole "thing" which could have been
anything, and in the failing test was a 1G guard region
- The whole wrong thing could be very large, resulting in OOM when
trying to write the minidump, which was the symptom of the bug.
Don't attempt to continue of SP isn't at least in a mapping, and don't
capture too-large regions for the stack as nothing useful can come of
that anyway.
New test added: ProcessSnapshotFuchsiaTest.InvalidStackPointer.
Bug: fuchsia:41212
Change-Id: Ifb48fd8b4b5b2f0cf10ab97e01dbd8b842368775
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/1912942
Commit-Queue: Scott Graham <scottmg@chromium.org>
Reviewed-by: Francois Rousseau <frousseau@google.com>
This allows partial reading of the current process (e.g. modules or
CrashpadInfo), even though the memory map read (and so thread retrieval)
will fail if ProcessSnapshotFuchsia is used on the current process.
This is a follow up to
db6f51d3fc
which broke the CrashpadInfoClientOptions.* tests.
Bug: fuchsia:34598
Change-Id: Ifa17b4dbefcd198ff67ecea91f946cfa2439ca4c
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/1776936
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
Commit-Queue: Scott Graham <scottmg@chromium.org>
From looking at some syzkaller logs, the earliest error seems to be
[...ERROR memory_map_fuchsia.cc:47] zx_object_get_info ZX_INFO_PROCESS_MAPS: ZX_ERR_BAD_STATE (-20)
which might be caused by a failure to suspend. That error causes
ProcessReaderFuchsia's memory_map_ Initialize() to fail, but that was
mistakenly being ignored. Later calls then fail as well.
While we're here, also check a couple other Initialize()s that were
sometimes triggering later on.
Bug: https://bugs.fuchsia.dev/p/fuchsia/issues/detail?id=34598
Change-Id: I9f9d7315f504e6ec69308add20e2737ce5c5f644
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/1763028
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
Commit-Queue: Scott Graham <scottmg@chromium.org>
Stack mappings can be enormous for some processes dwarfing all other
data and making the .dmp useless. It isn't useful to capture beyond the
stack pointer, so grab only from the stack base to the stack pointer.
In the default config (safestack enabled), this isn't a major problem.
However, Chromium has safestack disabled, along with a large stack size,
so dumps with many threads become very large.
Bug: https://bugs.fuchsia.dev/p/fuchsia/issues/detail?id=6425
Bug: chromium:821951
Change-Id: Iebefc5fe43e3d1bc4d8b66c107d3ab8ae5b3f68b
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/1758702
Commit-Queue: Scott Graham <scottmg@chromium.org>
Reviewed-by: Francois Rousseau <frousseau@google.com>
On Fuchsia, executables and loadable modules don't have a name at build
time so we use "<_>" as module name to index their symbols on the crash
server. We need to use the same dummy value at run time.
Bug: fuchsia/DX-1193
Tested: `fx run-test crashpad_test`
Change-Id: Ie926a6d26cb52679ccfac767db098c9fbfd21dd8
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/1548230
Commit-Queue: Francois Rousseau <frousseau@google.com>
Reviewed-by: Scott Graham <scottmg@chromium.org>
Test: Existing
Change-Id: I36fde186de372d2b86807f4da4e6e589a1b19706
Reviewed-on: https://chromium-review.googlesource.com/c/1395479
Reviewed-by: Scott Graham <scottmg@chromium.org>
Commit-Queue: Scott Graham <scottmg@chromium.org>
- Endian-swaps the 3 integer fields of the build id when returning it
for use as the module id (see bug 229).
- Removes the "app:" prefix on the main binary, as this prevents the
crash server from matching the binary name (and it isn't particularly
useful anyway)
- Map "<vDSO>" to "libzircon.so" as that's what it actually is, so that
symbols for it can be found.
Bug: crashpad:196, crashpad:229
Change-Id: Ie4abc732b7696345b96c34dbb1a7d2cc2cfcf77f
Reviewed-on: https://chromium-review.googlesource.com/1035461
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
Commit-Queue: Scott Graham <scottmg@chromium.org>
Conversion to CPUContext is currently only implemented for x64.
Bug: crashpad:196
Change-Id: I3fb8541f70a6f8d6f12c02e6b17c78e07e195056
Reviewed-on: https://chromium-review.googlesource.com/1007967
Commit-Queue: Scott Graham <scottmg@chromium.org>
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
This implementation has some limitations as documented in the header,
however, threads must be suspended in order to use the register capture
debug API so this is somewhat useful for now in the context of
generate_dump.
Also, refactor some child-object retrieval helpers used in a few places.
Bug: crashpad:196
Change-Id: I1fdae5fc3d4b43841e535724eac10c1e58af04c5
Reviewed-on: https://chromium-review.googlesource.com/1007966
Commit-Queue: Scott Graham <scottmg@chromium.org>
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
