]> git.openfabrics.org - ~emulex/infiniband.git/commitdiff
libertas: don't reset card on error when it is being removed
authorDaniel Drake <dsd@laptop.org>
Mon, 30 Jul 2012 21:58:04 +0000 (22:58 +0100)
committerJohn W. Linville <linville@tuxdriver.com>
Thu, 2 Aug 2012 17:51:47 +0000 (13:51 -0400)
On an OLPC XO-1.5 we have seen the following situation:
 - the system starts going into suspend
 - no wake params are set, so the mmc layer removes the card
 - during remove, we send a command to the card
 - that command fails, causing if_sdio's reset method to try and remove
   the mmc card in attempt to reset it
 - the mmc layer is not happy about being asked to remove a card that
   it is already removing, and the kernel crashes

While the MMC layer could possibly be taught to behave better here,
it also seems sensible for libertas not to try and reset a card if
we're in the process of removing it anyway.

Signed-off-by: Daniel Drake <dsd@laptop.org>
Acked-by: Dan Williams <dcbw@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
drivers/net/wireless/libertas/main.c

index 58048189bd24475f3ca18960822b6b18219e637c..fe1ea43c5149ef03e9ed0e2809695773fa2a576d 100644 (file)
@@ -571,7 +571,10 @@ static int lbs_thread(void *data)
                        netdev_info(dev, "Timeout submitting command 0x%04x\n",
                                    le16_to_cpu(cmdnode->cmdbuf->command));
                        lbs_complete_command(priv, cmdnode, -ETIMEDOUT);
-                       if (priv->reset_card)
+
+                       /* Reset card, but only when it isn't in the process
+                        * of being shutdown anyway. */
+                       if (!dev->dismantle && priv->reset_card)
                                priv->reset_card(priv);
                }
                priv->cmd_timed_out = 0;