This is the mail archive of the
mailing list for the GCC project.
Re: [PATCH]: R10000 Needs LL/SC Workaround in Gcc
On Tue, 4 Nov 2008, Kumba wrote:
> But I guess the question I'm pondering, is just how rare would it be for
> someone to actually need a MIPS-I binary with ll/sc and branch-likely fixes to
> run on something like an R10000? Rare enough to justify denying them that
> particular command argument combination, and thus taking Option #1? Or go the
> extra mile for Option #2? I don't know if that's my call to really make,
> since I lack the statistical data to know who would be affected, and in what
> ways (i.e., do they have alternative methods, such as MIPS-II, etc..).
Workarounds should be as cheap as possible maintenance-wise. My vote is
for requiring people in the need to use broken R10k revisions to choose a
compatible ISA. It actually makes sense to use 64-bit software on these
systems implying at least MIPS III, which is also another argument not to
try to bend backwards and support MIPS I with R10k workarounds.