Staging: dgnc: check kzalloc result
PATCH 1/1 Staging: dgnc: check kzalloc result
From: Matteo Semenzato <[email protected]>
Return -ENOMEM if allocating brd->flipbuf fails.
Signed-off-by Matteo Semenzato <[email protected]>
--- drivers/staging/dgnc/dgnc_driver.c | 4 ++++ 1 file changed, 4 insertions(+)
drivers/staging/dgnc/dgnc_driver.c +4 -0
@@ -622,6 +622,10 @@ static int dgnc_found_board(struct pci_dev *pdev, int id)* context, and there are no locks held. */ brd->flipbuf =kzalloc (MYFLIPLEN ,GFP_KERNEL ); +if (!brd->flipbuf) { +kfree (brd); +return -ENOMEM ; + }wake_up_interruptible (&brd->state_wait);
Signature · 1 line
2.3.1
Quoted 14 lines · Matteo Semenzato
From: Matteo Semenzato <[email protected]>
Return
-ENOMEMif allocatingbrd->flipbuffails.Signed-off-by Matteo Semenzato <[email protected]>--- drivers/staging/dgnc/dgnc_driver.c | 4 ++++ 1 file changed, 4 insertions(+)drivers/staging/dgnc/dgnc_driver.c +0
-0
drivers/staging/dgnc/dgnc_driver.c static int dgnc_found_board(struct pci_dev *pdev, int id) · line 627
Quoted 6 lines
drivers/staging/dgnc/dgnc_driver.c +3
-0@@ -622,6 +622,10 @@ static int dgnc_found_board(struct pci_dev *pdev, int id)* context, and there are no locks held. */ brd->flipbuf =kzalloc (MYFLIPLEN ,GFP_KERNEL ); +if (!brd->flipbuf) { +kfree (brd); +return -ENOMEM ;
This leaves a freed pointer in "dgnc_Board[dgnc_NumBoards]" which is sort of ugly. It doesn't free "brd->msgbuf_head" which is a bug.
Please read my essay on error hanling:
https://plus.google.com/106378716002406849458
What we want is :
ret = do_one(); if (ret) return ret; ret = do_two(); if (ret) goto undo_one; ret = do_three(); if (ret) goto undo_two; return 0;
err_undo_two:
undo_two(); err_undo_one: undo_one(); return ret;
Basically, a list of commands and a mirror list of undos. This is standard kernel style and the error handling basically writes itself automatically.
regards,
dan carpenter
Quoted 11 lines
drivers/staging/dgnc/dgnc_driver.c +1
-0+ }wake_up_interruptible (&brd->state_wait);Signature · 6 lines
2.3.1 _______________________________________________ devel mailing list [email protected] http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel