This is the mail archive of the
mailing list for the glibc project.
Re: [PATCH] Remove duplicate ifunc benchtests.
- From: Will Newton <will dot newton at linaro dot org>
- To: Siddhesh Poyarekar <siddhesh at redhat dot com>
- Cc: Ondřej Bílka <neleai at seznam dot cz>, libc-alpha <libc-alpha at sourceware dot org>
- Date: Tue, 26 Nov 2013 16:08:18 +0000
- Subject: Re: [PATCH] Remove duplicate ifunc benchtests.
- Authentication-results: sourceware.org; auth=none
- References: <20131112101625 dot GA23672 at domone dot podge> <20131125092759 dot GL19834 at spoyarek dot pnq dot redhat dot com>
On 25 November 2013 09:27, Siddhesh Poyarekar <firstname.lastname@example.org> wrote:
> On Tue, Nov 12, 2013 at 11:16:25AM +0100, OndÅej BÃlka wrote:
>> Current benchmarks print same data twice, when ifunc is not defined it
>> covers only some implementations and when it is defined then all
>> implementations are tested (results of ifunc/nonifunc may differ a lot
>> but that is a separate bug report.)
>> This can be solved by dropping a ifunc files and defining TEST_IFUNC
>> OK to commit?
> Looks OK to me.
>> diff --git a/benchtests/bench-string.h b/benchtests/bench-string.h
>> index 53e796a..46b4138 100644
>> --- a/benchtests/bench-string.h
>> +++ b/benchtests/bench-string.h
>> @@ -17,6 +17,7 @@
>> <http://www.gnu.org/licenses/>. */
>> #include <sys/cdefs.h>
>> +#define TEST_IFUNC 1
>> typedef struct
> Please post an additional patch to just remove TEST_IFUNC altogether,
> assuming it to be 1.
It also appears that benchtests/bench-strsep-ifunc.c was added to the
tree between this patch being created and committed.
Toolchain Working Group, Linaro