From: Jeremy S. <js...@mv...> - 2002-07-25 00:38:24
|
Hi folks, I'd like to apply the attached patches for a couple of problems I've noticed in recent days. (1) Sending signals via setup_rt_sigframe(): because __NR_rt_sigreturn is > 127, the attempt to put it into R3 for the trap like setup_sigframe() has a problem with sign-extension. Corrected by adding an extu.b instruction; copied same to setup_sigframe() for consistency. Also, call setup for second and third arguments is incorrect, putting them on stack rather than in r5/r6. The patch to arch/sh/kernel/signal.c attempts to fix these. (2) In sh_rtc_gettimeofday() there is some code to check that the register values are valid and reasonable; but in some cases I still get bogus values into xtime at init. The patch to arch/sh/kernel/rtc.c attempts to tighten these checks. Any objections to putting these in? Anything I missed regarding these? Thanks, --Jeremy |