On January 25, 2023 1:57:15 AM PST, Ammar Faizi <ammarfaizi2@xxxxxxxxxxx> wrote: >On Wed, Jan 25, 2023 at 12:39:26AM -0800, H. Peter Anvin wrote: >> > /* Set IP and CX to match so that SYSRET can happen. */ >> > ctx->uc_mcontext.gregs[REG_RIP] = rip; >> > ctx->uc_mcontext.gregs[REG_RCX] = rip; >> >> It would be interesting to have the syscall handler try both with and >> without this (so it would end up doing both IRET and SYSCALL on legacy.) >> Perhaps SIGUSR1 versus SIGUSR2... > >Just to clarify this more so I am sure I understand it correctly. > >Did you mean to have the same signal handler without modifiying >'REG_RCX' but still change 'REG_RIP'? > >IOW, we want to only *remove*: > > ctx->uc_mcontext.gregs[REG_RCX] = rip; > >and *keep*: > > ctx->uc_mcontext.gregs[REG_RIP] = rip; > >for the SIGUSR2 handler. Thus, inside the entry64 we will jump to the >iret path because %rcx != %r11 upon rt_sigreturn()? > I guess it would depend on what they "normally" are. My #1 impulse would be to leave them both unchanged.
