|
From: Carl E. L. <ce...@li...> - 2014-02-19 20:34:30
|
IBM will be supporting both big endian and little endian on power. I am
working on porting Valgrind to run under little endian. I am getting a
number of errors as follows that do not occur when run on big endian
Valgrind:
Conditional jump or move depends on uninitialised value(s)
Use of uninitialised value of size 8
I suspect that there is still an issue with the setting and/or checking
of the Vbits in memcheck that determine if the values are initialized or
not.
In file memchec/mc_machine.c there is a comment before function
get_otrack_shadow_offset_wrk() about the function being endian dependent
"This function is dependent on the host's endianness, hence we
assert that the use case is supported. "
I am not seeing where the endian dependency is and specifically how I
should change the code to go from PPC64 big endian to PPC little endian.
Hope someone can help identify where the endian dependency is and how to
fix it for power. I am still trying to figure out what all this code is
trying to do. Thanks.
Carl Love
|
|
From: Dejan J. <dej...@rt...> - 2014-02-20 10:43:46
|
Hello Carl, mips32 and mips64 are both supporting big and little endian. Maybe you could check: #if defined (_MIPSEL) // for little and #if defined (_MIPSEB) // for big endian. Regards, Dejan On 19.02.2014. 21:34, Carl E. Love wrote: > > IBM will be supporting both big endian and little endian on power. I am > working on porting Valgrind to run under little endian. I am getting a > number of errors as follows that do not occur when run on big endian > Valgrind: > > Conditional jump or move depends on uninitialised value(s) > Use of uninitialised value of size 8 > > I suspect that there is still an issue with the setting and/or checking > of the Vbits in memcheck that determine if the values are initialized or > not. > > In file memchec/mc_machine.c there is a comment before function > get_otrack_shadow_offset_wrk() about the function being endian dependent > > "This function is dependent on the host's endianness, hence we > assert that the use case is supported. " > > I am not seeing where the endian dependency is and specifically how I > should change the code to go from PPC64 big endian to PPC little endian. > Hope someone can help identify where the endian dependency is and how to > fix it for power. I am still trying to figure out what all this code is > trying to do. Thanks. > > Carl Love > > > ------------------------------------------------------------------------------ > Managing the Performance of Cloud-Based Applications > Take advantage of what the Cloud has to offer - Avoid Common Pitfalls. > Read the Whitepaper. > http://pubads.g.doubleclick.net/gampad/clk?id=121054471&iu=/4140/ostg.clktrk > _______________________________________________ > Valgrind-developers mailing list > Val...@li... > https://lists.sourceforge.net/lists/listinfo/valgrind-developers > |