From mboxrd@z Thu Jan 1 00:00:00 1970 From: Michael Schmitz Subject: Re: [PATCH 1/2] alpha/ptrace: Record and handle the absence of switch_stack Date: Mon, 21 Jun 2021 16:08:35 +1200 Message-ID: <700d8acd-d3df-a026-8b1d-55eeae836eea@gmail.com> References: <87sg1p30a1.fsf@disp2133> <87pmwsytb3.fsf@disp2133> <87sg1lwhvm.fsf@disp2133> <6e47eff8-d0a4-8390-1222-e975bfbf3a65@gmail.com> <924ec53c-2fd9-2e1c-bbb1-3fda49809be4@gmail.com> <87eed4v2dc.fsf@disp2133> <5929e116-fa61-b211-342a-c706dcb834ca@gmail.com> <87fsxjorgs.fsf@disp2133> <87zgvqor7d.fsf_-_@disp2133> <87mtrpg47k.fsf@disp2133> <87pmwlek8d.fsf_-_@disp2133> <87k0mtek4n.fsf_-_@disp2133> <393c37de-5edf-effc-3d06-d7e63f34a317@gmail.com> <60c0fe00-b966-6385-d348-f6dd45277113@gmail.com> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Return-path: DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20161025; h=subject:to:references:cc:from:message-id:date:user-agent :mime-version:in-reply-to:content-transfer-encoding; bh=F9odJSZJ3JHPCDq5xWJRw+RMGbd9Q+9zPppzBJyCxj4=; b=Eho+Hv3biN4uuSiZMxo0AjUNa9n+ziDPIuyDrm4TVlHFw2x+yxKEQi7R8uZS5S9XNk gGgEAY96ZGN7wRhP53985TAMlMuEZfbZLbo2EkhHMls4CW8p7PdmYz1Cai5kkoswAZ9k gYBovh0PHzAkx9wudrQVZ0L01Ha8zSx6asccS+ynHXLe519RNSWSkK/OS03dv0NUgSy2 z3CES8B12c2YuTbQMdxiyKGHKbvEt7uOdZGxNuQi6aD0jEIYM/7NhB/xYLnZHkm//IAI T8Wg9Z4swG5YS40bp0Ki/q5lMWliWOe+1KDUR8mH/5KcL0dLqru00c2OduyIF3kp8yXw 0bog== In-Reply-To: List-ID: Content-Type: text/plain; charset="us-ascii"; format="flowed" To: Linus Torvalds Cc: "Eric W. Biederman" , linux-arch , Jens Axboe , Oleg Nesterov , Al Viro , Linux Kernel Mailing List , Richard Henderson , Ivan Kokshaysky , Matt Turner , alpha , Geert Uytterhoeven , linux-m68k , Arnd Bergmann , Ley Foon Tan , Tejun Heo , Kees Cook Hi Linus, I realized that the patch is still incomplete when answering Al... Am 21.06.2021 um 15:37 schrieb Linus Torvalds: > On Sun, Jun 20, 2021 at 8:18 PM Michael Schmitz wrote: >> >> I hope that makes more sense? > > So the problem is in your debug patch: you don't set that > TIS_SWITCH_STACK in nearly enough places. > > In this particular example, I think it's that you don't set it in > do_trace_exit, so when you strace the process, the system call exit - > which is where the return value will be picked up - gets that warning. > > You did set TIS_SWITCH_STACK on trace_entry, but then it's cleared > again during the system call, and not set at the trace_exit path. > Oddly, your debug patch also _clears_ it on the exit path, but it > doesn't set it when do_trace_exit does the SAVE_SWITCH_STACK. > > You oddly also set it for __sys_exit, but not all the other special > system calls that also do that SAVE_SWITCH_STACK. That's the one I used to test whether my debug patch had any ill side effects (i.e. smashing the stack) late yesterday. Forgot to add that to the other cases. > > Really, pretty much every single case of SAVE_SWITCH_STACK would need > to set it. Not just do_trace_enter/exit Yes - done that now and the warning is gone. > It's why I didn't like Eric's debug patch either. It's quite expensive > to do, partly because you look up that curptr thing. All very nasty. I need to talk to Geert and Andreas to find where register a1 is preserved, but if I have to reload a1 all the time, this won't be useful except for debugging. > It would be *much* better to make the flag be part of the stack frame, > but sadly at least on alpha we had exported the format of that stack > frame to user space. Same on m68k, but can we push a flag _after_ the switch stack? > Anyway, I think these debug patches are not just expensive but the > m68k one most definitely is also very incomplete. Yes, I've seen that in the meantime. Need to triple check my work next time. Sorry for the extra noise! Cheers, Michael > > Linus >