RESOLVED FIXED 112400
[iOS] Update StringImpl's equal to have a single version on all supported Apple CPUs 
https://bugs.webkit.org/show_bug.cgi?id=112400
Summary [iOS] Update StringImpl's equal to have a single version on all supported App...
Benjamin Poulain
Reported 2013-03-14 20:22:46 PDT
[iOS] Update StringImpl's equal to have a single version on all supported Apple CPUs
Attachments
Patch (4.54 KB, patch)
2013-03-14 20:32 PDT, Benjamin Poulain 		no flags
DetailsFormatted DiffDiff
Patch (4.57 KB, patch)
2013-03-15 12:36 PDT, Benjamin Poulain 		msaboff: review+
DetailsFormatted DiffDiff
Show Obsolete (1) View All Add attachment proposed patch, testcase, etc.
Benjamin Poulain
Comment 1 2013-03-14 20:32:27 PDT
Created attachment 193225 [details] Patch
Benjamin Poulain
Comment 2 2013-03-14 20:38:29 PDT
<rdar://problem/12437844>
Benjamin Poulain
Comment 3 2013-03-15 12:36:35 PDT
Created attachment 193354 [details] Patch
Michael Saboff
Comment 4 2013-03-15 13:47:19 PDT
Comment on attachment 193354 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=193354&action=review I'd prefer a little more descriptive labels, but that predates this patch. > Source/WTF/ChangeLog:15 > + each iteration and work in the negative space for the tail. This free one register typo, should be "frees" > Source/WTF/ChangeLog:18 > + negative space, we can test for the Carry flag to jumb back to the next LDR. typo, should be "jump"
Benjamin Poulain
Comment 5 2013-03-15 15:13:13 PDT
Committed r145949: <http://trac.webkit.org/changeset/145949>
Benjamin Poulain
Comment 6 2013-03-15 15:14:34 PDT
I ended up keeping the labels. For some reason, I still got linking error with %= once compiling in release.
Benjamin Poulain
Comment 7 2013-03-15 15:18:18 PDT
For anyone interested in using the same code. You can use this on any hardware with: -It supports unaligned LDR and LDRH (in hardware, otherwise there is no point). -The A bit in the CP15 C1 Control Register must be Clear.
Adam Barth
Comment 8 2013-03-19 12:55:06 PDT
Is there a benchmark that shows this patch is a win? It seems sad to have a PLATFORM(IOS) ifdef in StringImpl.h.
Note You need to log in before you can comment on or make changes to this bug.