On 10.04.2008 [15:55:08 -0700], Nishanth Aravamudan wrote:
> On 10.04.2008 [16:50:22 -0500], Jon Tollefson wrote:
> > This patch moves the check of the return value from gethugepagesize() to a 
> > common function, check_hugepagesize(),
> > rather then repeating it in each test.  gethugepagesize() is updated to set 
> > errno depending on the reason for the failure.
> > 
> > Only tests that were already calling gethugepagesize() are calling 
> > check_hugepagesize().  Attached also is the output of the test suite run 
> > with 16G pages on POWER before this change and after.
> > 
> > Changes since v4
> >     -save errno so that the same value can be returned on every call to 
> > gethugepagesize()
> >     -removed check_hugepagesize() call in linkhuge and linkshare since the 
> > return will be checked in morecore and elflink in part 2 of this patch
> >     -removed the call_gethugepagesize() function since linkhuge and 
> > linkshare no longer need it
> > 
> > Changes since v3
> >     -incorporated Mel Gorman's function that uses dlopen so that tests 
> > don't have to be linked against libhugetlbfs
> >     -used suggestions from Nishanth Aravamudan to use a static inline 
> > function and rename to check_hugepagesize()
> >     -added check_hugepagesize() call to linkhuge and linkshare tests
> > 
> > 
> > Signed-off-by: Jon Tollefson <[EMAIL PROTECTED]>
> <snip>
> > xB.linkhuge (32):       libhugetlbfs: ERROR: Hugepage size too large
> > Failed to map hugepage segment 0: 20000000-20000000 (errno=22)
> > ./run_tests.sh: line 34: 17526 Aborted                 
> > PATH="obj$BITS:$PATH" LD_LIBRARY_PATH="$LD_LIBRARY_PATH:../obj$BITS" $ENV 
> > "$@"
> 
> Hrm, the message from libhugetlbfs is good, but why are we continuing
> the mmap() still? We should just fail out from there...

Oh I see. Your second patch didn't fix up the error handling in the
other call to gethugepagesize() in elflink.c Will fix it locally.

Thanks,
Nish

-- 
Nishanth Aravamudan <[EMAIL PROTECTED]>
IBM Linux Technology Center

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
Don't miss this year's exciting event. There's still time to save $100. 
Use priority code J8TL2D2. 
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
Libhugetlbfs-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libhugetlbfs-devel

Reply via email to