Add '-fno-delete-null-pointer-checks' to gcc CFLAGS
authorEugene Teo <eteo@redhat.com>
Wed, 15 Jul 2009 06:59:10 +0000 (14:59 +0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Thu, 16 Jul 2009 16:19:16 +0000 (09:19 -0700)
commita3ca86aea507904148870946d599e07a340b39bf
treeb5d6c35f8048bf7f071f1fc16f5d543c07a725b2
parent4a21b8cb3550f19f838f7c48345fbbf6a0e8536b
Add '-fno-delete-null-pointer-checks' to gcc CFLAGS

Turning on this flag could prevent the compiler from optimising away
some "useless" checks for null pointers.  Such bugs can sometimes become
exploitable at compile time because of the -O2 optimisation.

See http://gcc.gnu.org/onlinedocs/gcc-4.1.2/gcc/Optimize-Options.html

An example that clearly shows this 'problem' is commit 6bf67672.

 static void __devexit agnx_pci_remove(struct pci_dev *pdev)
 {
     struct ieee80211_hw *dev = pci_get_drvdata(pdev);
-    struct agnx_priv *priv = dev->priv;
+    struct agnx_priv *priv;
     AGNX_TRACE;

     if (!dev)
         return;
+    priv = dev->priv;

By reverting this patch, and compile it with and without
-fno-delete-null-pointer-checks flag, we can see that the check for dev
is compiled away.

    call    printk  #
-   testq   %r12, %r12  # dev
-   je  .L94    #,
    movq    %r12, %rdi  # dev,

Clearly the 'fix' is to stop using dev before it is tested, but building
with -fno-delete-null-pointer-checks flag at least makes it harder to
abuse.

Signed-off-by: Eugene Teo <eugeneteo@kernel.sg>
Acked-by: Eric Paris <eparis@redhat.com>
Acked-by: Wang Cong <amwang@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Makefile