rio: typo in bitwise AND expression.
Subsystems (2)
PATCH 1/1 rio: typo in bitwise AND expression.
Hi Rogier,
here's a patch to fix a typo in rio_linux which affects both kernel 2.4 and 2.6. It's not big deal it seems as it only affects the irq-less path.
I found this one like that :
$ grep -r '[^&]&[^&]*![^=]' drivers/char/
I'm sure others will find more efficient rules to catch such errors.
Regards,
Willy
Quoted 1 line
From
4fb85842b76ad28893ea2aeaeb6dbc4e3f5a2deeMon Sep 17 00:00:00 2001
From: Willy Tarreau <[email protected]>
Date: Wed, 22 Nov 2006 23:54:48 +0100
Subject: [PATCH] rio: typo in bitwise AND expression.
The line :
hp->Mode &= !RIO_PCI_INT_ENABLE;
is obviously wrong as RIO_PCI_INT_ENABLE=0x04 and is used as a bitmask 2 lines before. Getting no IRQ would not disable RIO_PCI_INT_ENABLE but rather RIO_PCI_BOOT_FROM_RAM which equals 0x01.
Obvious fix is to change ! for ~.
--- drivers/char/rio/rio_linux.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)
drivers/char/rio/rio_linux.c +1 -1
@@ -1143,7 +1143,7 @@ #endif /* PCI */rio_dprintk (RIO_DEBUG_INIT ,"Enabling interrupts on rio card.\n" ); hp->Mode |=RIO_PCI_INT_ENABLE ; }else - hp->Mode &= !RIO_PCI_INT_ENABLE ; + hp->Mode &= ~RIO_PCI_INT_ENABLE ;rio_dprintk (RIO_DEBUG_INIT ,"New Mode: %x\n" , hp->Mode);rio_start_card_running (hp); }
Signature · 1 line
1.4.2.4
On Wed, Nov 22, 2006 at 11:58:56PM +0100, Willy Tarreau wrote:
Seems ok.
Patrick
Quoted 58 lines
Hi Rogier,
here's a patch to fix a typo in
rio_linuxwhich affects both kernel 2.4 and 2.6. It's not big deal it seems as it only affects the irq-less path.I found this one like that :
$ grep -r '[^&]&[^&]*![^=]' drivers/char/I'm sure others will find more efficient rules to catch such errors.
Regards,
WillyFrom
4fb85842b76ad28893ea2aeaeb6dbc4e3f5a2deeMon Sep 17 00:00:00 2001
From: Willy Tarreau <[email protected]>
Date: Wed, 22 Nov 2006 23:54:48 +0100
Subject: [PATCH] rio: typo in bitwise AND expression.The line :
hp->Mode &= !RIO_PCI_INT_ENABLE;is obviously wrong as RIO_PCI_INT_ENABLE=0x04 and is used as a bitmask 2 lines before. Getting no IRQ would not disable
RIO_PCI_INT_ENABLEbut ratherRIO_PCI_BOOT_FROM_RAMwhich equals0x01.Obvious fix is to change ! for ~.
Signed-off-by Willy Tarreau <[email protected]>--- drivers/char/rio/rio_linux.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)drivers/char/rio/rio_linux.c +1
-1@@ -1143,7 +1143,7 @@ #endif /* PCI */rio_dprintk (RIO_DEBUG_INIT ,"Enabling interrupts on rio card.\n" ); hp->Mode |=RIO_PCI_INT_ENABLE ; }else - hp->Mode &= !RIO_PCI_INT_ENABLE ; + hp->Mode &= ~RIO_PCI_INT_ENABLE ;rio_dprintk (RIO_DEBUG_INIT ,"New Mode: %x\n" , hp->Mode);rio_start_card_running (hp); }Signature · 7 lines
1.4.2.4 - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Willy Tarreau
AppliedApplied to 2.4, thanks Patrick.
Quoted 66 lines · Patrick vd Lageweg
On Wed, Nov 22, 2006 at 11:58:56PM +0100, Willy Tarreau wrote:
Seems ok.
Signed-off-by Patrick vd Lageweg <[email protected]>Patrick
Quoted 58 lines
Hi Rogier,
here's a patch to fix a typo in
rio_linuxwhich affects both kernel 2.4 and 2.6. It's not big deal it seems as it only affects the irq-less path.I found this one like that :
$ grep -r '[^&]&[^&]*![^=]' drivers/char/I'm sure others will find more efficient rules to catch such errors.
Regards,
WillyFrom
4fb85842b76ad28893ea2aeaeb6dbc4e3f5a2deeMon Sep 17 00:00:00 2001
From: Willy Tarreau <[email protected]>
Date: Wed, 22 Nov 2006 23:54:48 +0100
Subject: [PATCH] rio: typo in bitwise AND expression.The line :
hp->Mode &= !RIO_PCI_INT_ENABLE;is obviously wrong as RIO_PCI_INT_ENABLE=0x04 and is used as a bitmask 2 lines before. Getting no IRQ would not disable
RIO_PCI_INT_ENABLEbut ratherRIO_PCI_BOOT_FROM_RAMwhich equals0x01.Obvious fix is to change ! for ~.
Signed-off-by Willy Tarreau <[email protected]>--- drivers/char/rio/rio_linux.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)drivers/char/rio/rio_linux.c +1
-1@@ -1143,7 +1143,7 @@ #endif /* PCI */rio_dprintk (RIO_DEBUG_INIT ,"Enabling interrupts on rio card.\n" ); hp->Mode |=RIO_PCI_INT_ENABLE ; }else - hp->Mode &= !RIO_PCI_INT_ENABLE ; + hp->Mode &= ~RIO_PCI_INT_ENABLE ;rio_dprintk (RIO_DEBUG_INIT ,"New Mode: %x\n" , hp->Mode);rio_start_card_running (hp); }Signature · 7 lines
1.4.2.4 - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
PATCH 1/1 6pack: fix "&= !" typo
In reply to Willy TarreauAndreas, is this correct?
---------------------------------SIXP_RX_DCD_MASK is 0x18, so the command below will make cmd 0 always. This is likely wrong.
---
drivers/net/hamradio/6pack.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/hamradio/6pack.c +1 -1
@@ -914,7 +914,7 @@ static void decode_prio_command(struct sprintk (KERN_DEBUG "6pack: protocol violation\n" ); else sp->status = 0; - cmd &= !SIXP_RX_DCD_MASK ; + cmd &= ~SIXP_RX_DCD_MASK ; } sp->status = cmd &SIXP_PRIO_DATA_MASK ; }else {/* output watchdog char if idle */
Re: [PATCH] 6pack: fix "&= !" typo
Quoted 6 lines · Alexey Dobriyan
Andreas, is this correct?
---------------------------------SIXP_RX_DCD_MASKis0x18, so the command below will make cmd 0 always. This is likely wrong.Signed-off-by Alexey Dobriyan <[email protected]>
This one is already merged.
It's funny how long this bug survived - it's in the kernel since the 6pack driver was first merged that is 2.1 or 2.2 ...
Ralf
Re: [PATCH] 6pack: fix "&= !" typo
Quoted 7 lines · Ralf Baechle
Quoted 5 lines · Alexey Dobriyan
Andreas, is this correct?
---------------------------------SIXP_RX_DCD_MASKis0x18, so the command below will make cmd 0 always. This is likely wrong.
Quoted 6 lines
Quoted 1 line
Signed-off-by Alexey Dobriyan <[email protected]>This one is already merged.
It's funny how long this bug survived - it's in the kernel since the 6pack driver was first merged that is 2.1 or 2.2 ...
One more reason to perform more code reviews helped with automated tools. We found this one and the rio's one while discussing with Jean Delvare about such bugs, and firing a random grep to illustrate how easy it could be to spot bugs similar to Alexey's "&&" instead of "&" ...
I think that we should at least take a look at all lines in the pre-processed code having both '!' and '&' on the same line. There are a lot of them, but divided by a sufficient number of volunteers, we might catch a bunch of them.
BTW, has anyone a good idea on how to make gcc dump the preprocessed files for everything it builds ? I mean, just by changing some variables in the Makefile.
Quoted 1 line
Ralf
Cheers,
Willy
Re: [PATCH] 6pack: fix "&= !" typo
Quoted 5 lines · Willy Tarreau
One more reason to perform more code reviews helped with automated tools. We found this one and the rio's one while discussing with Jean Delvare about such bugs, and firing a random grep to illustrate how easy it could be to spot bugs similar to Alexey's "&&" instead of "&" ...
Quoted 7 lines
I think that we should at least take a look at all lines in the pre-processed code having both '!' and '&' on the same line. There are a lot of them, but divided by a sufficient number of volunteers, we might catch a bunch of them.
BTW, has anyone a good idea on how to make gcc dump the preprocessed files for everything it builds ? I mean, just by changing some variables in the
Makefile.
It seems to might be about to re-invent sparse which is probably the right framework for such semantic tests.
Ralf