make gendev_rel_sem a compat_semaphore

Hi Ingo,

I was getting the following:

BUG: nonzero lock count 10 at exit time?
        modprobe: 2972 [ffff81007e1aaf70, 116]

Call Trace:<ffffffff8014e2db>{printk_task+43} <ffffffff8015040f>{check_no_held_locks+111}
       <ffffffff80136d3c>{do_exit+3036} <ffffffff80136f5c>{do_group_exit+268}
       <ffffffff80136f72>{sys_exit_group+18} <ffffffff8011e471>{ia32_sysret+0}

---------------------------
| preempt count: 00000000 ]
| 0-level deep critical section nesting:
----------------------------------------
hdc: ATAPI 40X DVD-ROM DVD-R CD-R/RW drive, 2048kB Cache, UDMA(33) Uniform CD-ROM driver Revision: 3.20

BUG: modprobe/2972, lock held at task exit time!
 [ffffffff8809fd00] {(struct semaphore *)(&hwif->gendev_rel_sem)}

.. held by: modprobe: 2972 [ffff81007e1aaf70, 116]
... acquired at: init_hwif_data+0xaf/0x1a0 [ide_core]

[snipped to not be so annoying]

Looking into this I see that gendev_rel_sem, which is only used when the device is unregistered, is defined as a semaphore. This patch changes this to be a compat_semaphore.

-- Steve

Index: linux-2.6.14-rt13/include/linux/ide.h
===================================================================

include/linux/ide.h +1 -1

@@ -910,7 +910,7 @@
 	unsigned	sg_mapped  : 1;	/* sg_table and sg_nents are ready */
 
 	struct device	gendev;
-	struct semaphore gendev_rel_sem; /* To deal with device release() */
+	struct compat_semaphore gendev_rel_sem; /* To deal with device release() */
 
 	void		*hwif_data;	/* extra hwif data */
 

I was getting the following:

BUG: nonzero lock count 10 at exit time?
        modprobe: 2972 [ffff81007e1aaf70, 116]

Looking into this I see that gendev_rel_sem, which is only used when the device is unregistered, is defined as a semaphore. This patch changes this to be a compat_semaphore.

Crap! I sent this with my kihontech email. Please respond to this instead. I'm still in the process of moving to my new machine, and the email was messed up.

Thanks,

-- Steve

Quoted 37 lines · Steven Rostedt

Hi Ingo,

I was getting the following:

BUG: nonzero lock count 10 at exit time?
        modprobe: 2972 [ffff81007e1aaf70, 116]

Call Trace:<ffffffff8014e2db>{printk_task+43} <ffffffff8015040f>{check_no_held_locks+111}
       <ffffffff80136d3c>{do_exit+3036} <ffffffff80136f5c>{do_group_exit+268}
       <ffffffff80136f72>{sys_exit_group+18} <ffffffff8011e471>{ia32_sysret+0}

---------------------------
| preempt count: 00000000 ]
| 0-level deep critical section nesting:
----------------------------------------
hdc: ATAPI 40X DVD-ROM DVD-R CD-R/RW drive, 2048kB Cache, UDMA(33) Uniform CD-ROM driver Revision: 3.20

BUG: modprobe/2972, lock held at task exit time!
 [ffffffff8809fd00] {(struct semaphore *)(&hwif->gendev_rel_sem)}

.. held by: modprobe: 2972 [ffff81007e1aaf70, 116]
... acquired at: init_hwif_data+0xaf/0x1a0 [ide_core]

[snipped to not be so annoying]

Looking into this I see that gendev_rel_sem, which is only used when the device is unregistered, is defined as a semaphore. This patch changes this to be a compat_semaphore.

-- Steve

Index: linux-2.6.14-rt13/include/linux/ide.h
===================================================================

include/linux/ide.h line 916

Quoted 8 lines

include/linux/ide.h +1 -1

@@ -910,7 +910,7 @@
 	unsigned	sg_mapped  : 1;	/* sg_table and sg_nents are ready */
 
 	struct device	gendev;
-	struct semaphore gendev_rel_sem; /* To deal with device release() */
+	struct compat_semaphore gendev_rel_sem; /* To deal with device release() */
 
 	void		*hwif_data;	/* extra hwif data */
 

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

How would you feel if we move this lock to a completion() . It's such a small lock, and completions are faster anyway .. It might be a good test bed for some other compat locks .