Please do not reply to this email. Use the web interface provided at: http://bugs.ecos.sourceware.org/show_bug.cgi?id=1001607
Ilija Kocho <[email protected]> changed: What |Removed |Added ---------------------------------------------------------------------------- Attachment #1881|0 |1 is obsolete| | --- Comment #23 from Ilija Kocho <[email protected]> 2012-08-09 11:11:57 BST --- Created an attachment (id=1882) --> (http://bugs.ecos.sourceware.org/attachment.cgi?id=1882) Enter default VSR from Usagefault VSR if not FPU exception. 120809 (In reply to comment #22) > (In reply to comment #19) > > (In reply to comment #17) > > Looking at the patch you posted in comment #21, do you need to save r2/r3 at > all? In fact since hal_deliver_usagefault_exception doesn't use its 'regs' > argument, should we drop that and only be saving, I guess, r12 and lr? Although > we have to stash r0 into r4 as before, so that too. All the other registers > should be unaffected by hal_deliver_usagefault_exception, since unlike other > user exception handling by default_exception_vsr, we know what this handler > does. And basepri should also be unaffected I would have thought. You are right, now that exception's job is predetermined, it can be optimized. We don't have to save all registers, indeed, and r12 can be omitted (auto-saved by exception), also maybe lr - but we need to keep stack aligned to 8 bytes anyway. [snip] > I think with this route then hal_usagefault_exception_vsr and > hal_deliver_usagefault_exception should be renamed to > hal_fpu_usagefault_exception_vsr and hal_deliver_fpu_usagefault_exception > instead. > I renamed hal_deliver_usagefault..., but technically hal_usagefault_exception_vsr still handles all usagefault exceptions, though some of them with help of hal_default_exception_vsr I am going to implement other changes for which which I think that are clear, and for others I will reply in another post. Поздрав Илија -- Configure bugmail: http://bugs.ecos.sourceware.org/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
