Summary: | [MIPS] Fix build on MIPS32r1 | ||||||
---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | Dominik Inführ <dominik.infuehr> | ||||
Component: | New Bugs | Assignee: | Nobody <webkit-unassigned> | ||||
Status: | RESOLVED FIXED | ||||||
Severity: | Normal | CC: | achristensen, benjamin, cdumez, cmarcelo, commit-queue, dbates, ews-watchlist, fpizlo, ggaren, jiewen_tan, keith_miller, loic.yhuel, mark.lam, msaboff, olivier.blin, saam, webkit-bug-importer, ysuzuki | ||||
Priority: | P2 | Keywords: | InRadar | ||||
Version: | WebKit Nightly Build | ||||||
Hardware: | Unspecified | ||||||
OS: | Unspecified | ||||||
Attachments: |
|
Description
Dominik Inführ
2018-05-24 06:10:21 PDT
Created attachment 341185 [details]
Patch
This looks ok, and builds fine. Here are the different cases I can think of : - -march=mips32 -mfp32 : mtc1/mfc1, OK - -march=mips32 -mfpxx (so __mips_isa_rev == 1, __mips_fpr == 0) : "as" would fail with "Error: float register should be even, was xx" (the only way would be to an intermediate memory transfer). Nothing changes here, so I think no one uses this configuration so it's fine if WebKit doesn't support it. - -march=mips32r2 -mfp32 : mfhc1 and mthc1 are used even if not necessary, but this isn't an issue - -march=mips32r2 -mfpxx : mfhc1 and mthc1, OK - -mfp64 : gcc already ensures mfhc1 and mthc1 are available I confirm it builds fine on a MIPS32r1 target, and jsc runs correctly too. Comment on attachment 341185 [details]
Patch
r=me
Comment on attachment 341185 [details] Patch Clearing flags on attachment: 341185 Committed r232290: <https://trac.webkit.org/changeset/232290> All reviewed patches have been landed. Closing bug. |