Skip to contentSkip to conversation

"gdth: update deprecated pci_find_device" is incorrect

DISCUSSION Jiri Slaby 1 participant 4 messages Feb 14, 2008 SCSI

lore
Share & tools
lore

Jiri Slaby

author
#0 lore raw

Hi,

commit
99109301d103fbf0de43fc5a580a406c12a501e0
in jejb/scsi-rc-fixes-2.6.git is incorrect. You don't decrement pci refcount on exit. Also you do not so on fail paths... I wonder why these mistakes happen every second time somebody tries to do such change.

It leaked into mainline yet after "whole" two days, but what exactly drives me crazy is, that Jeff commented it in similar way and nobody reflected it!

author
#1 lore raw
Quoted 7 lines · Jiri Slaby

Hi,

commit
99109301d103fbf0de43fc5a580a406c12a501e0
in jejb/scsi-rc-fixes-2.6.git is incorrect. You don't decrement pci refcount on exit. Also you do not so on fail paths... I wonder why these mistakes happen every second time somebody tries to do such change.

It leaked into mainline yet after "whole" two days, but what exactly drives me crazy is, that Jeff commented it in similar way and nobody reflected it!

BTW if you have more than one card, you protected the driver from no race, since you don't pci_dev_get of successfully grabbed cards.

author
#2 lore raw
Quoted 8 lines · Jiri Slaby
Quoted 7 lines · Jiri Slaby

Hi,

commit
99109301d103fbf0de43fc5a580a406c12a501e0
in jejb/scsi-rc-fixes-2.6.git is incorrect. You don't decrement pci refcount on exit. Also you do not so on fail paths... I wonder why these mistakes happen every second time somebody tries to do such change.

It leaked into mainline yet after "whole" two days, but what exactly drives me crazy is, that Jeff commented it in similar way and nobody reflected it!

Yeah, you did (I read the thread on wrong server obviusly), sorry...

author
#3 lore raw
In reply to Jiri Slaby
Quoted 4 lines · Jiri Slaby
Quoted 3 lines · Jiri Slaby

commit
99109301d103fbf0de43fc5a580a406c12a501e0
in jejb/scsi-rc-fixes-2.6.git is incorrect. You don't decrement pci

[...]

Quoted 1 line

BTW if you have more than one card, you protected the driver from no race, since you don't pci_dev_get of successfully grabbed cards.

(this still holds)