Closed Bug 853128 Opened 12 years ago Closed 12 years ago

TEST-UNEXPECTED-FAIL | WindowsDllInterceptor | Failed to hook NtFlushBuffersFile from ntdll.dll

Categories

(Firefox Build System :: General, defect)

x86_64
Windows 8
defect
Not set
normal

Tracking

(Not tracked)

RESOLVED FIXED
mozilla23

People

(Reporter: speciesx, Assigned: m_kato)

References

Details

(Keywords: intermittent-failure)

Attachments

(1 file, 1 obsolete file)

https://tbpl.mozilla.org/php/getParsedLog.php?id=20890142&tree=Firefox TEST-UNEXPECTED-FAIL | WindowsDllInterceptor | Failed to hook NtFlushBuffersFile from ntdll.dll e:\builds\moz2_slave\m-cen-w64-pgo-0000000000000000\build\obj-firefox\toolkit\xre\test\win\Makefile:41:0: command '../../../../dist/bin/TestDllInterceptor.exe' failed, return code 1 e:\builds\moz2_slave\m-cen-w64-pgo-0000000000000000\build\config\rules.mk:1781:0: command 'C:/mozilla-build/python27/python.exe e:/builds/moz2_slave/m-cen-w64-pgo-0000000000000000/build/build/pymake/pymake/../make.py -C win check' failed, return code 2 e:\builds\moz2_slave\m-cen-w64-pgo-0000000000000000\build\config\rules.mk:1781:0: command 'C:/mozilla-build/python27/python.exe e:/builds/moz2_slave/m-cen-w64-pgo-0000000000000000/build/build/pymake/pymake/../make.py -C test check' failed, return code 2 e:\builds\moz2_slave\m-cen-w64-pgo-0000000000000000\build\config\rules.mk:1781:0: command 'C:/mozilla-build/python27/python.exe e:/builds/moz2_slave/m-cen-w64-pgo-0000000000000000/build/build/pymake/pymake/../make.py -C xre check' failed, return code 2 e:\builds\moz2_slave\m-cen-w64-pgo-0000000000000000\build\config\rules.mk:1781:0: command 'C:/mozilla-build/python27/python.exe e:/builds/moz2_slave/m-cen-w64-pgo-0000000000000000/build/build/pymake/pymake/../make.py -C toolkit check' failed, return code 2
not need NtFlushbufferFile hook. But we keep this (see bug 850957).
Attachment #740145 - Flags: review?(ehsan)
Attachment #740145 - Attachment description: turn off injecting NtFlushBuffersFile on x64 → turn off injecting test of NtFlushBuffersFile on x64
Comment on attachment 740145 [details] [diff] [review] turn off injecting test of NtFlushBuffersFile on x64 oops. other patch is merged
Attachment #740145 - Flags: review?(ehsan)
Attachment #740145 - Attachment is obsolete: true
Attachment #740146 - Flags: review?(ehsan)
Considering bug 848020 removed the hooking of that function, you can just remove the test completely.
Blocks: 848020
(In reply to Mike Hommey [:glandium] from comment #10) > Considering bug 848020 removed the hooking of that function, you can just > remove the test completely. Ehsan says, "I'd rather us keep this test" at bug 850957 comment #2.
(In reply to Makoto Kato from comment #11) > (In reply to Mike Hommey [:glandium] from comment #10) > > Considering bug 848020 removed the hooking of that function, you can just > > remove the test completely. > > Ehsan says, "I'd rather us keep this test" at bug 850957 comment #2. But the function is not hooked anymore. https://hg.mozilla.org/mozilla-central/rev/fe8a8cad48a4
(In reply to Mike Hommey [:glandium] from comment #12) > (In reply to Makoto Kato from comment #11) > > (In reply to Mike Hommey [:glandium] from comment #10) > > > Considering bug 848020 removed the hooking of that function, you can just > > > remove the test completely. > > > > Ehsan says, "I'd rather us keep this test" at bug 850957 comment #2. > > But the function is not hooked anymore. > https://hg.mozilla.org/mozilla-central/rev/fe8a8cad48a4 So I make a proposal removing it by first patch of bug 850957, but ehsan wants to keep this. (bug 850957 comment #2) by code review. First patch of bug 850957 was just remove this API hook test.
(In reply to Mike Hommey [:glandium] from comment #12) > (In reply to Makoto Kato from comment #11) > > (In reply to Mike Hommey [:glandium] from comment #10) > > > Considering bug 848020 removed the hooking of that function, you can just > > > remove the test completely. > > > > Ehsan says, "I'd rather us keep this test" at bug 850957 comment #2. > > But the function is not hooked anymore. > https://hg.mozilla.org/mozilla-central/rev/fe8a8cad48a4 That doesn't matter, this test is testing hooking up functions with those specific opcodes at the beginning, which is something that we want to keep working in case someone needs to hook such a function in the future.
Attachment #740146 - Flags: review?(ehsan) → review+
Assignee: nobody → m_kato
Status: UNCONFIRMED → RESOLVED
Closed: 12 years ago
Resolution: --- → FIXED
"Failed to hook NtWriteFile from ntdll.dll" is also showing up; should we just disable this one as well? We do seem to hook it in PoisonWrite(), so maybe it needs to be fixed?
(In reply to comment #19) > "Failed to hook NtWriteFile from ntdll.dll" is also showing up; should we just > disable this one as well? We do seem to hook it in PoisonWrite(), so maybe it > needs to be fixed? Can you please file a follow-up about that? This could be bad.
(In reply to :Ehsan Akhgari (needinfo? me!) from comment #20) > (In reply to comment #19) > > "Failed to hook NtWriteFile from ntdll.dll" is also showing up; should we just > > disable this one as well? We do seem to hook it in PoisonWrite(), so maybe it > > needs to be fixed? > > Can you please file a follow-up about that? This could be bad. Ah, we need alalyze more mov opcode and syscall. 64bit implementation of NtWriteFile simply call to system call of kernel.
By the way, if we could at the same time the test is failing dump the first x bytes of the function, that would avoid having to look them up manually each time there's a new failure. x=16 should be enough for both x86 and x86-64.
filed bug 870218.
Product: Core → Firefox Build System
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: