Hi Linus,
I found a SMP race in execve():
cpu1:
void main() { execve(argv[0],argv,envp);}
cpu2:
$ cd /proc/<pid>;while true;do cat cmdline>/dev/null;done
After a few seconds, execve on cpu1 will fail: there is a race between
concurrent access_process_vm and setup_arg_pages().
setup_arg_pages() must acquire the mmap semaphore, that solved the
problem.
Could you add this patch to the next kernel?
[tested with 2.3.99-pre8, i386, SMP]
<<<<<<<<<<<
--- 2.3/fs/exec.c Fri May 12 08:52:15 2000
+++ build-2.3/fs/exec.c Sat May 13 21:24:38 2000
@@ -288,6 +288,7 @@
if (!mpnt)
return -ENOMEM;
+ down(¤t->mm->mmap_sem);
{
mpnt->vm_mm = current->mm;
mpnt->vm_start = PAGE_MASK & (unsigned long) bprm->p;
@@ -311,6 +312,7 @@
}
stack_base += PAGE_SIZE;
}
+ up(¤t->mm->mmap_sem);
return 0;
}
>>>>>>>>>>>
-- Manfred- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Mon May 15 2000 - 21:00:23 EST