From [email protected] on August 04, 2010 08:51:28

part2

from drmemory.

From [email protected] on August 06, 2010 07:02:36

I've added a couple of debug printfs locally to investigate this issue.
The assert happens when err->count==2 and ->errtype=0 which stands for ERROR_UNADDRESSABLE

from drmemory.

From [email protected] on August 06, 2010 07:33:31

Re comment #2:
This has happened with another suppression file:

UNADDRESSABLE ACCESS
ntdll.dll!towlower
ntdll.dll!KiUserExceptionDispatcher
*!DebuggerProbe
*!DebuggerKnownHandle

LoadLibrary calls give some strange reports

UNADDRESSABLE ACCESS
ntdll.dll!strchr
KERNEL32.dll!LoadLibraryExW

https://code.google.com/p/drmemory/issues/detail?id=11 UNADDRESSABLE ACCESS

*!_chkstk

and err->suppressed=1

from drmemory.

From [email protected] on August 06, 2010 07:59:15

hmmm, probably the assert should just be removed: a suppressed error w/ multiple instances will legitimately have id==0 and count>1

from drmemory.

From [email protected] on August 06, 2010 11:19:30

This is partially fixed in r29 , I'll continue investigation if other crashes are still present

from drmemory.

From [email protected] on August 09, 2010 08:13:20

The assert is still firing if
a) suppression file is present
b) the number of reports exceeds -report_max value
c) some of the reports match the suppressions

The default -report_max value is 20000 so probably that's why you didn't see it happening. However, on Chromium tests there are so many reports (most of them are false) that this number is easily exceeded.

Reproducer:
code snippet and suppression file from issue #14 ,
tested on Dr. Memory r30 / dynamorio r383 :
$ drmemory.exe -report_max 5 -suppress supp.txt -- test.exe
...
ASSERT FAILURE (thread 90812): drmemory\report.c:1088: err->id != 0 (duplicate should have id)

from drmemory.

From [email protected] on August 10, 2010 02:08:10

Hopefully, this is fixed by r31

Status: Fixed

from drmemory.