List:Commits« Previous MessageNext Message »
From:Vladislav Vaintroub Date:March 5 2008 10:34pm
Subject:RE: bk commit into 5.1 tree (tsmith:1.2543) BUG#34297
View as plain text  
Hi Tim,
Unfortunately it does look as fix by suspicion, I see no evidence of a compiler bug.
Unless someone really  looked into disassembly, proved the compiler backed was wrong and
filed a bug report at compiler vendor, I do not consider the problem to be solved. Maybe
someone already did, but then I miss a bug reference number attached to the changeset
comment. And I'm not suggesting you should do this, the bug happens with Innodb only, so
they are probably the right address (and hey, sure Oracle got some kind of
Golden-Premium-Whatever paid Microsoft support).

Ok, my point is: if one suspects compiler bug in a relatively new compiler (I'm not
speaking about old ones, for which fixes  hardly possible), then there should be a
reference to bug id filed for compiler manufacturer. If we do not do this, their bugs
will never ever be fixed.

Cheers,
Vlad

> -----Original Message-----
> From: tim@stripped [mailto:tim@stripped]
> Sent: Wednesday, March 05, 2008 9:27 PM
> To: commits@stripped
> Subject: bk commit into 5.1 tree (tsmith:1.2543) BUG#34297
> 
> Below is the list of changes that have just been committed into a local
> 5.1 repository of tsmith.  When tsmith does a push these changes
> will be propagated to the main repository and, within 24 hours after
> the
> push, to the public repository.
> For information on how to access the public repository
> see http://dev.mysql.com/doc/mysql/en/installing-source-tree.html
> 
> ChangeSet@stripped, 2008-03-05 13:27:17-07:00, tsmith@stripped
> +1 -0
>   Bug #34297: MySQL Server crashes when processing large table
> 
>   Compiler bug.  Fix is to reduce optimization if the compiler is
>   "Visual Studio 8 2005 Win64".  This was already done in 5.0 as a
>   fix for Bug 19424.  The present bug is 5.1+ only.
> 
>   storage/innobase/CMakeLists.txt@stripped, 2008-03-05 13:27:10-07:00,
> tsmith@stripped +7 -0
>     Conditionally set COMPILE_FLAGS -Od on mem/mem0mem.c and
>     mem/mem0pool.c, to work around a compiler bug in VS 8 2005 Win64.
> 
> diff -Nrup a/storage/innobase/CMakeLists.txt
> b/storage/innobase/CMakeLists.txt
> --- a/storage/innobase/CMakeLists.txt	2008-01-29 19:58:55 -07:00
> +++ b/storage/innobase/CMakeLists.txt	2008-03-05 13:27:10 -07:00
> @@ -17,6 +17,13 @@ SET(CMAKE_CXX_FLAGS_DEBUG "${CMAKE_CXX_F
>  SET(CMAKE_C_FLAGS_DEBUG "${CMAKE_C_FLAGS_DEBUG} -DSAFEMALLOC -
> DSAFE_MUTEX")
>  ADD_DEFINITIONS(-DMYSQL_SERVER -D_WIN32 -D_LIB)
> 
> +# Bug#19424 - InnoDB: Possibly a memory overrun of the buffer being
> freed (64-bit Visual C)
> +# Removing Win64 compiler optimizations for all innodb/mem/* files.
> +IF(CMAKE_GENERATOR MATCHES "Visual Studio 8 2005 Win64")
> +	SET_SOURCE_FILES_PROPERTIES(mem/mem0mem.c mem/mem0pool.c
> +							 PROPERTIES COMPILE_FLAGS -Od)
> +ENDIF(CMAKE_GENERATOR MATCHES "Visual Studio 8 2005 Win64")
> +
>  INCLUDE_DIRECTORIES(${CMAKE_SOURCE_DIR}/include
> ${CMAKE_SOURCE_DIR}/zlib
>                      ${CMAKE_SOURCE_DIR}/storage/innobase/include
>                      ${CMAKE_SOURCE_DIR}/storage/innobase/handler
> 
> --
> MySQL Code Commits Mailing List
> For list archives: http://lists.mysql.com/commits
> To unsubscribe:
> http://lists.mysql.com/commits?unsub=1


Thread
bk commit into 5.1 tree (tsmith:1.2543) BUG#34297tim5 Mar
  • RE: bk commit into 5.1 tree (tsmith:1.2543) BUG#34297Vladislav Vaintroub5 Mar