[Bug libc/357] New: getcontext() on ppc32 destroys saved parameter 1 in caller's frame

[Bug libc/357] New: getcontext() on ppc32 destroys saved parameter 1 in caller's frame

[qboosh at pld-linux dot org]

GLIBC_2.3.4 getcontext() on ppc32 changes the value of the first
parameter passed to caller's function.
Testcase:

$ cat tst_getcontext.c
#include <ucontext.h>

void test(volatile int a, volatile int b)
{
        ucontext_t uc;
        printf("before: %d %d\n", a, b);
        getcontext(&uc);
        printf("after : %d %d\n", a, b);
}

int main()
{
        test(1, 2);
}
$ ./tst_getcontext
before: 1 2
after : 2147478448 2

("volatile" is not needed when tst_getcontext is compiled without any
optimization)

It's caused by the first instruction of __getcontext()
(sysdeps/unix/sysv/linux/powerpc/powerpc32/getcontext.S:29):

ENTRY(__getcontext)
        stw     r3,_FRAME_PARM_SAVE1(r1)

__getcontext doesn't have its own stack frame, so it uses caller's
stack frame, thus destroying it...

The same applies to GLIBC_2.3.4 swapcontext(), but it may be less/not(?)
harmful as successful swapcontext() never returns.

-- 
           Summary: getcontext() on ppc32 destroys saved parameter 1 in
                    caller's frame
           Product: glibc
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P2
         Component: libc
        AssignedTo: gotom at debian dot or dot jp
        ReportedBy: qboosh at pld-linux dot org
                CC: glibc-bugs at sources dot redhat dot com
  GCC host triplet: ppc-linux


http://sources.redhat.com/bugzilla/show_bug.cgi?id=357

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/357] getcontext() on ppc32 destroys saved parameter 1 in caller's frame

[sjmunroe at us dot ibm dot com]

------- Additional Comments From sjmunroe at us dot ibm dot com  2004-08-30 23:22 -------
Created an attachment (id=179)
 --> (http://sources.redhat.com/bugzilla/attachment.cgi?id=179&action=view)
Here is a patch to try

This patch fixes the problem in the testcase and preserves all registers except
R3. It was the minimum change that would work. But Paul Mackerras points out I
don't really have to perserve any of the volatile registers and burning a few
more registers would speed the code up a bit. I'll look at that tomorrow.

-- 


http://sources.redhat.com/bugzilla/show_bug.cgi?id=357

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/357] getcontext() on ppc32 destroys saved parameter 1 in caller's frame

[aj at suse dot de]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|gotom at debian dot or dot  |sjmunroe at us dot ibm dot
                   |jp                          |com
             Status|NEW                         |ASSIGNED


http://sources.redhat.com/bugzilla/show_bug.cgi?id=357

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/357] getcontext() on ppc32 destroys saved parameter 1 in caller's frame

[qboosh at pld-linux dot org]

------- Additional Comments From qboosh at pld-linux dot org  2004-08-31 21:07 -------
Patch fixes problem for me (both in attached testcase and real-world application
on which I discovered it).

-- 


http://sources.redhat.com/bugzilla/show_bug.cgi?id=357

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/357] getcontext() on ppc32 destroys saved parameter 1 in caller's frame

[sjmunroe at us dot ibm dot com]

------- Additional Comments From sjmunroe at us dot ibm dot com  2004-09-02 18:44 -------
Created an attachment (id=183)
 --> (http://sources.redhat.com/bugzilla/attachment.cgi?id=183&action=view)
Final patch with testcase update

Since I have to stack a frame anyway burning more volatiles does not buy much.
So I started with the privious patch and improved the scheduling. 

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
 Attachment #179 is|0                           |1
           obsolete|                            |


http://sources.redhat.com/bugzilla/show_bug.cgi?id=357

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug libc/357] getcontext() on ppc32 destroys saved parameter 1 in caller's frame

[jakub at redhat dot com]

------- Additional Comments From jakub at redhat dot com  2004-09-03 08:09 -------
Should be fixed in CVS.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED


http://sources.redhat.com/bugzilla/show_bug.cgi?id=357

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.