blk: Check if_type in blk_get_devnum_by_typename
PATCH 1/1 blk: Check if_type in blk_get_devnum_by_typename
While searching for a BLK device, this function checks only for a matching devnum. It should check if_type, too.
---
drivers/block/blk-uclass.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/block/blk-uclass.c +1 -1
@@ -112,7 +112,7 @@ struct blk_desc *blk_get_devnum_by_typename(const char *if_typename, int devnum)debug ("%s: if_type=%d, devnum=%d: %s, %d, %d\n" , __func__, if_type, devnum, dev->name, desc->if_type, desc->devnum); -if (desc->devnum != devnum) +if (desc->if_type != if_type || desc->devnum != devnum)continue ;/* Find out the parent device uclass */
Signature · 1 line
2.20.1.98.gecbdaf0899
PATCH 1/1 blk: Check if_type in blk_get_devnum_by_typename
Quoted 2 lines · Juha Sarlin
While searching for a BLK device, this function checks only for a matching devnum. It should check
if_type, too.
Could you, please, describe in which cases you have observed a problem and how it can be reproduced.
According to the function description the relevant interface type check is device_get_uclass_id(dev->parent) != uclass_id
drivers/block/blk-uclass.c struct blk_desc *blk_get_devnum_by_typename(const char *if_typename, int devnum)
Quoted 12 lines
Signed-off-by Juha Sarlin <[email protected]>---
drivers/block/blk-uclass.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/block/blk-uclass.c +0
-0@@ -112,7 +112,7 @@ struct blk_desc *blk_get_devnum_by_typename(const char *if_typename, int devnum)
Should the logic need changing, please, update the function description. Please, use Sphinx style for the function description.
Best regards
Heinrich
Quoted 9 lines
drivers/block/blk-uclass.c +1
-1debug ("%s: if_type=%d, devnum=%d: %s, %d, %d\n" , __func__, if_type, devnum, dev->name, desc->if_type, desc->devnum); -if (desc->devnum != devnum) +if (desc->if_type != if_type || desc->devnum != devnum)continue ;/* Find out the parent device uclass */
PATCH 1/1 blk: Check if_type in blk_get_devnum_by_typename
Quoted 12 lines · Heinrich Schuchardt
Quoted 2 lines · Juha Sarlin
While searching for a BLK device, this function checks only for a matching devnum. It should check
if_type, too.Could you, please, describe in which cases you have observed a problem and how it can be reproduced.
According to the function description the relevant interface type check is
device_get_uclass_id(dev->parent)!=uclass_id
I was wrong, it isn't really a bug. I was misled by all the other blk-finding functions that check if_type instead of parent class. I think that checking if_type would work here, too.
Or perhaps this case is the first step towards removing the if_type field in the future? There seems to be a 1-1 mapping from almost every if_type to uclass_id.
PATCH 1/1 blk: Check if_type in blk_get_devnum_by_typename
Hi Juha,
Quoted 11 lines · Juha Sarlin
Quoted 11 lines · Heinrich Schuchardt
Quoted 2 lines · Juha Sarlin
While searching for a BLK device, this function checks only for a matching devnum. It should check
if_type, too.Could you, please, describe in which cases you have observed a problem and how it can be reproduced.
Quoted 7 lines
Quoted 2 lines
According to the function description the relevant interface type check is
device_get_uclass_id(dev->parent)!=uclass_idI was wrong, it isn't really a bug. I was misled by all the other blk-finding functions that check
if_typeinstead of parent class. I think that checkingif_typewould work here, too.Or perhaps this case is the first step towards removing the
if_typefield in the future? There seems to be a 1-1 mapping from almost everyif_typetouclass_id.
Thanks for the patch. In this case it is the intended behaviour I think, because if_type_to_uclass_id() gives us the uclass to use, and we check that immediately below your patch. Since there is a 1:1 correspondence between if_type and uclass_id (apart from those we want to remove) we don't need to check both.
Yes it would be nice to remove if_type one day.
Regards,
Simon