Re: Segfault in _Unwind_* code called from pthread_exit

From: Tijl Coosemans <>
Date: Sun, 29 Oct 2017 18:23:51 +0100
On Sat, 26 Aug 2017 21:40:34 +0300 Konstantin Belousov <> wrote:
> On Sat, Aug 26, 2017 at 08:28:13PM +0200, Tijl Coosemans wrote:
>> I did consider using
>> a CFI directive (see patch below) and it works, but it's architecture
>> specific and it's inserted after the function prologue so there's still
>> a window of a few instructions where a stack unwinder will try to use
>> the return address.
>> Index: lib/libthr/thread/thr_create.c
>> ===================================================================
>> --- lib/libthr/thread/thr_create.c      (revision 322802)
>> +++ lib/libthr/thread/thr_create.c      (working copy)
>> _at__at_ -251,6 +251,7 _at__at_ create_stack(struct pthread_attr *pattr)
>>  static void
>>  thread_start(struct pthread *curthread)
>>  {
>> +       __asm(".cfi_undefined %rip");
>>         sigset_t set;
>>         if (curthread->attr.suspend == THR_CREATE_SUSPENDED)  
> I like this approach much more than the previous patch.  What can be
> done is to provide asm trampoline which calls thread_start().  There you
> can add the .cfi_undefined right at the entry.
> It is somewhat more work than just setting the return address on the
> kernel-constructed pseudo stack frame, but I believe this is ultimately
> correct way.  You still can do it only on some arches, if you do not
> have incentive to code asm for all of them.

Ok, but then there are two ways to implement the trampoline:

	movq $0,(%rsp)
	jmp thread_start
	subq $8,%rsp
	call thread_start

With 1) you're setting the return address to zero anyway, so you might
as well do that in the kernel like my first patch.  With 2) you're
setting up a new call frame, basically redoing what the kernel already
did and on i386 this also means copying the function argument.

Do you have any preference (or better alternatives), because I think I
still prefer my first patch.  It's the caller's job to set up the call
frame, in this case the kernel.  And if the kernel handles it then it
also works with (hypothetical) implementations other than libthr.

> Also crt1 probably should get the same treatment, despite we already set
> %rbp to zero AFAIR.

I haven't checked but I imagine the return address of the process entry
point is always zero because the stack is all zeros.
Received on Sun Oct 29 2017 - 16:25:07 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:41:13 UTC