Skip to contentSkip to conversation

fix adm9240 oops

PATCH New Jonathan Corbet 3 participants 5 messages Aug 25, 2005 – Aug 26, 2005

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 fix adm9240 oops

The adm9240 driver, in adm9240_detect(), allocates a structure. The error path attempts to kfree() a subfield of that structure, resulting in an oops (or slab corruption) if the hardware is not present. This one seems worth fixing for 2.6.13.

jon

drivers/hwmon/adm9240.c renamed from drivers/hwmon/adm9240.c.orig +1 -1
@@ -616,7 +616,7 @@ static int adm9240_detect(struct i2c_ada
 
 	return 0;
 exit_free:
-	kfree(new_client);
+	kfree(data);
 exit:
 	return err;
 }
#1 lore raw

PATCH 1/1 drivers/hwmon/*: kfree() correct pointers

The adm9240 driver, in adm9240_detect(), allocates a structure. The error path attempts to kfree() ->client field of it (second one), resulting in an oops (or slab corruption) if the hardware is not present.

->client field in adm1026, adm1031, smsc47b397 and smsc47m1 is the first in ${HWMON}_data structure, but fix them too.

---

 drivers/hwmon/adm1026.c    |    2 +-
 drivers/hwmon/adm1031.c    |    2 +-
 drivers/hwmon/adm9240.c    |    2 +-
 drivers/hwmon/smsc47b397.c |    2 +-
 drivers/hwmon/smsc47m1.c   |    2 +-
 5 files changed, 5 insertions(+), 5 deletions(-)
drivers/hwmon/adm1026.c +1 -1
diff -uprN linux-vanilla/drivers/hwmon/adm1026.c linux-hwmon/drivers/hwmon/adm1026.c
@@ -1691,7 +1691,7 @@ int adm1026_detect(struct i2c_adapter *a
 
 	/* Error out and cleanup code */
 exitfree:
-	kfree(new_client);
+	kfree(data);
 exit:
 	return err;
 }
drivers/hwmon/adm1031.c +1 -1
diff -uprN linux-vanilla/drivers/hwmon/adm1031.c linux-hwmon/drivers/hwmon/adm1031.c
@@ -834,7 +834,7 @@ static int adm1031_detect(struct i2c_ada
 	return 0;
 
 exit_free:
-	kfree(new_client);
+	kfree(data);
 exit:
 	return err;
 }
drivers/hwmon/adm9240.c +1 -1
diff -uprN linux-vanilla/drivers/hwmon/adm9240.c linux-hwmon/drivers/hwmon/adm9240.c
@@ -616,7 +616,7 @@ static int adm9240_detect(struct i2c_ada
 
 	return 0;
 exit_free:
-	kfree(new_client);
+	kfree(data);
 exit:
 	return err;
 }
drivers/hwmon/smsc47b397.c +1 -1
diff -uprN linux-vanilla/drivers/hwmon/smsc47b397.c linux-hwmon/drivers/hwmon/smsc47b397.c
@@ -298,7 +298,7 @@ static int smsc47b397_detect(struct i2c_
 	return 0;
 
 error_free:
-	kfree(new_client);
+	kfree(data);
 error_release:
 	release_region(addr, SMSC_EXTENT);
 	return err;
drivers/hwmon/smsc47m1.c +1 -1
diff -uprN linux-vanilla/drivers/hwmon/smsc47m1.c linux-hwmon/drivers/hwmon/smsc47m1.c
@@ -495,7 +495,7 @@ static int smsc47m1_detect(struct i2c_ad
 	return 0;
 
 error_free:
-	kfree(new_client);
+	kfree(data);
 error_release:
 	release_region(address, SMSC_EXTENT);
 	return err;
#2 lore raw

Re: [PATCH] drivers/hwmon/*: kfree() correct pointers

Hi Alexey,

Quoted 7 lines

The adm9240 driver, in adm9240_detect(), allocates a structure. The error path attempts to kfree() ->client field of it (second one), resulting in an oops (or slab corruption) if the hardware is not present.

->client field in adm1026, adm1031, smsc47b397 and smsc47m1 is the first in ${HWMON}_data structure, but fix them too.

Already fixed in Greg's i2c tree and -mm for quite some time now...

http://www.kernel.org/pub/linux/kernel/people/gregkh/gregkh-2.6/gregkh-02-i2c/i2c-hwmon-class-01.patch

Thanks,

Signature · 1 line
Jean Delvare
#3 lore raw

Re: [PATCH] drivers/hwmon/*: kfree() correct pointers

Quoted 1 line

Already fixed in Greg's i2c tree and -mm for quite some time now...

So it is. The comment says, however, that "the existing code works somewhat by accident." In the case of the 9240 driver, however, the existing code demonstrably does not work - it oopsed on me. The patch in Greg's tree looks fine (it's a straightforward fix, after all); I'd recommend that it be merged before 2.6.13.

jon

Jonathan Corbet
Executive editor, LWN.net
[email protected]

#4 lore raw

Re: [PATCH] drivers/hwmon/*: kfree() correct pointers

Hi Jonathan,

Quoted 5 lines
Quoted 1 line

Already fixed in Greg's i2c tree and -mm for quite some time now...

So it is. The comment says, however, that "the existing code works somewhat by accident." In the case of the 9240 driver, however, the existing code demonstrably does not work - it oopsed on me.

I too did notice that the adm9240 case was worse than the four other ones back then, but when I tried to get it to crash, it never did. This is the reason why I did not push this patch upstream faster. I wonder why it now does oops on you.

I also believe that this patch was somewhat misnamed. It is not related to the new hwmon class, but jut happened to change the same part of these five drivers. With a better name, the patch would most probably have been selected by Greg in the last batch of i2c patches to Linus.

Quoted 2 lines

The patch in Greg's tree looks fine (it's a straightforward fix, after all);

I wouldn't call it straightforward, but it certainly has been reviewed and tested well enough by now to be considered safe.

Quoted 1 line

I'd recommend that it be merged before 2.6.13.

Fine with me.

Thanks,

Signature · 1 line
Jean Delvare