remove check_region in drivers-char-specialix.c
PATCH 1/1 remove check_region in drivers-char-specialix.c
Hi Andrew,
This is also a pretty simple case. We remove the wrapper and make sx__request_io_range return struct resource *. We check its value accordingly in the probing routine. It compiles cleanly here.
drivers/char/specialix.c renamed from drivers/char/specialix.c.orig +3 -10
@@ -338,15 +338,9 @@ static inline void sx_wait_CCR_off(struc* specialix IO8+ IO range functions. */ -static inline int sx_check_io_range (struct specialix_board * bp) +static inline struct resource *sx_request_io_range (struct specialix_board * bp) { -return check_region (bp->base,SX_IO_SPACE ); -} - - -static inline void sx_request_io_range (struct specialix_board * bp) -{ -request_region (bp->base, +return request_region (bp->base, bp->flags&SX_BOARD_IS_PCI ?SX_PCI_IO_SPACE :SX_IO_SPACE ,"specialix IO8+" ); }
@@ -495,7 +489,7 @@ static int sx_probe(struct specialix_boafunc_enter (); -if (sx_check_io_range (bp)) { +if (!sx_request_io_range (bp)) {func_exit ();return 1 ; }
@@ -583,7 +577,6 @@ static int sx_probe(struct specialix_boareturn -EIO ; } -sx_request_io_range (bp); bp->flags |=SX_BOARD_PRESENT ;/* Chip revcode pkgtype
___________________________________________________________
Gesendet von Yahoo! Mail - Jetzt mit 1GB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de
Quoted 5 lines · Borislav Petkov
Hi Andrew,
This is also a pretty simple case. We remove the wrapper and make
sx__request_io_rangereturn struct resource *. We check its value accordingly in the probing routine. It compiles cleanly here.
NAK. You've just introduced a pile of leaks - if sx_probe() fails after that call, you end up with region claimed and not released.
Quoted 9 lines · Al Viro
Quoted 5 lines · Borislav Petkov
Hi Andrew,
This is also a pretty simple case. We remove the wrapper and make
sx__request_io_rangereturn struct resource *. We check its value accordingly in the probing routine. It compiles cleanly here.NAK. You've just introduced a pile of leaks - if
sx_probe()fails after that call, you end up with region claimed and not released.
Andrew told me already today that Jeff[1] had sent a patch fixing all that. To prevent the leaks he's calling sx_release_io_range(bp) in every check before exiting sx_probe so this seems correct. A small question though: After calling sx_request_io_range() in the if-statement on line 499 is it ok to call sx_request_io_range() for a second time in a row on line 587? I think in this case the second call has to go, no?
[1]rsync://rsync.kernel.org/pub/scm/linux/kernel/git/jgarzik/misc-2.6.git
___________________________________________________________
Gesendet von Yahoo! Mail - Jetzt mit 1GB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de
Quoted 8 lines · Borislav Petkov
Andrew told me already today that Jeff[1] had sent a patch fixing all that. To prevent the leaks he's calling
sx_release_io_range(bp)in every check before exitingsx_probeso this seems correct. A small question though: After callingsx_request_io_range()in the if-statement on line 499 is it ok to callsx_request_io_range()for a second time in a row on line 587? I think in this case the second call has to go, no?[1]rsync://rsync.kernel.org/pub/scm/linux/kernel/git/jgarzik/misc-2.6.git
Huh? I don't see any specialix patches in that repository right now...
But yes, after successful request_region() you shouldn't call it again - that would simply fail.
Quoted 9 lines · Al Viro
Quoted 7 lines · Borislav Petkov
Andrew told me already today that Jeff[1] had sent a patch fixing all that. To prevent the leaks he's calling
sx_release_io_range(bp)in every check before exitingsx_probeso this seems correct. A small question though: After callingsx_request_io_range()in the if-statement on line 499 is it ok to callsx_request_io_range()for a second time in a row on line 587? I think in this case the second call has to go, no?
Quoted 3 lines
Quoted 1 line
[1]rsync://rsync.kernel.org/pub/scm/linux/kernel/git/jgarzik/misc-2.6.git
Huh? I don't see any specialix patches in that repository right now...
Quoted 1 line · Andrew Morton
Ewww... A lot of chunks consisting only of whitespace removals - great way to make patch less readable...
And yes, that second call of sx_request_io_range() must die. BTW, what's wrong with use of mdelay() instead of that sx_long_delay() junk? Replacing both calls of sx_long_delay() with mdelay(50) would do it...
Quoted 10 lines · Al Viro
Quoted 1 line · Andrew Morton
Ewww... A lot of chunks consisting only of whitespace removals - great way to make patch less readable...
And yes, that second call of
sx_request_io_range()must die. BTW, what's wrong with use ofmdelay()instead of thatsx_long_delay()junk? Replacing both calls ofsx_long_delay()with mdelay(50) would do it...
Trust me: mdelay didn't exist when I wrote that.
The code calls the private function that does what I think should've been kernel infrastructure all along to make it easy to either:
change the body of the function to call the new infrastructure, or replace the call of the private function with the call to the new infrastructure.
Roger.
Signature · 5 lines
** [email protected] ** http://www.BitWizard.nl/ ** +31-15-2600998 ** *-- BitWizard writes Linux device drivers for any device you may have! --* Q: It doesn't work. A: Look buddy, doesn't work is an ambiguous statement. Does it sit on the couch all day? Is it unemployed? Please be specific! Define 'it' and what it isn't doing. --------- Adapted from lxrbot FAQ
Quoted 10 lines · Al Viro
Quoted 1 line · Andrew Morton
Ewww... A lot of chunks consisting only of whitespace removals - great way to make patch less readable...
And yes, that second call of
sx_request_io_range()must die. BTW, what's wrong with use ofmdelay()instead of thatsx_long_delay()junk? Replacing both calls ofsx_long_delay()with mdelay(50) would do it...
There's a proper check_region removal for specialix.c patch from me queued up in the kernel-janitors tree. Including removal of the silly wrappers.