by Hi Linus, Dave,
This is basically a patch from Randolph Chung who tells me that when
a syscall is done from the kernel, you cannot pass user mode pointers
to it on some architectures. So we need to copy the sock_filter
array into kernel space before we pass it to the real system call.
His original patch (which does the same thing) has been tested on
parisc64-linux and this patch has been cross compiled for ppc64-linux.
Please apply.
--
Cheers,
http://www.canb.auug.org.au/~sfr/
diff -ruN 2.5.66-033114/net/compat.c 2.5.66-033114-netfix/net/compat.c
--- 2.5.66-033114/net/compat.c 2003-03-25 12:08:29.000000000 +1100
struct sock_fprog kfprog;
mm_segment_t old_fs;
compat_uptr_t uptr;
+ unsigned int fsize;
int ret;
__get_user(uptr, &fprog32->filter))
return -EFAULT;
- kfprog.filter = compat_ptr(uptr);
- /*
- * Since struct sock_filter is architecure independent,
- * we can just do the access_ok check and pass the
- * same pointer to the real syscall.
- */
- if (!access_ok(VERIFY_READ, kfprog.filter,
- kfprog.len * sizeof(struct sock_filter)))
+ fsize = kfprog.len * sizeof(struct sock_filter);
+ kfprog.filter = (struct sock_filter *)kmalloc(fsize, GFP_KERNEL);
+ if (kfprog.filter == NULL)
+ return -ENOMEM;
+ if (copy_from_user(kfprog.filter, compat_ptr(uptr), fsize)) {
+ kfree(kfprog.filter);
return -EFAULT;
+ }
old_fs = get_fs();
(char *)&kfprog, sizeof(kfprog));
set_fs(old_fs);
+ kfree(kfprog.filter);
return ret;
}
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/