usbnet: allow type check of devdbg arguments in non-debug build
Subsystems (4)
PATCH 1/1 usbnet: allow type check of devdbg arguments in non-debug build
improve usbnet's devdbg to always type-check diagnostic arguments, like dev_dbg (device.h). This makes no change to the resulting size of usbnet modules.
--- include/linux/usb/usbnet.h | 4 +++- 1 files changed, 3 insertions(+), 1 deletions(-)
include/linux/usb/usbnet.h +3 -1
@@ -197,7 +197,9 @@ extern int usbnet_nway_reset(struct net_device *net); #define devdbg(usbnet, fmt, arg...) \ printk(KERN_DEBUG "%s: " fmt "\n" , (usbnet)->net->name , ## arg)#else -#define devdbg(usbnet, fmt, arg...) do {} while(0) +#define devdbg(usbnet, fmt, arg...) \ + ({ if (0) printk(KERN_DEBUG "%s: " fmt "\n" , (usbnet)->net->name , \ + ## arg); 0; })#endif #define deverr(usbnet, fmt, arg...) \
Signature · 1 line
1.5.5.1
Jeff Garzik
AppliedQuoted 14 lines · Steve Glendinning
improve usbnet's devdbg to always type-check diagnostic arguments, like
dev_dbg(device.h). This makes no change to the resulting size of usbnet modules.Signed-off-by Steve Glendinning <[email protected]>--- include/linux/usb/usbnet.h | 4 +++- 1 files changed, 3 insertions(+), 1 deletions(-)include/linux/usb/usbnet.h +0
-0
include/linux/usb/usbnet.h extern int usbnet_nway_reset(struct net_device *net); · line 203
Quoted 8 lines
include/linux/usb/usbnet.h +3
-1@@ -197,7 +197,9 @@ extern int usbnet_nway_reset(struct net_device *net); #define devdbg(usbnet, fmt, arg...) \ printk(KERN_DEBUG "%s: " fmt "\n" , (usbnet)->net->name , ## arg)#else -#define devdbg(usbnet, fmt, arg...) do {} while(0) +#define devdbg(usbnet, fmt, arg...) \ + ({ if (0) printk(KERN_DEBUG "%s: " fmt "\n" , (usbnet)->net->name , \ + ## arg); 0; })#endif
applied
PATCH 1/1 usbnet: allow type check of devdbg arguments in non-debug build
In reply to Steve GlendinningFrom: Steve Glendinning <[email protected]>
Subject: usbnet: allow type check of devdbg arguments in non-debug build
Improve usbnet's devdbg to always type-check diagnostic arguments, like dev_dbg (device.h). This makes no change to the resulting size of usbnet modules.
---
[ refreshed against 2.6.29-rc1 GIT ]
include/linux/usb/usbnet.h | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
include/linux/usb/usbnet.h +3 -1
@@ -197,7 +197,9 @@ extern int usbnet_nway_reset(struct net_ #define devdbg(usbnet, fmt, arg...) \ printk(KERN_DEBUG "%s: " fmt "\n" , (usbnet)->net->name , ## arg)#else -#define devdbg(usbnet, fmt, arg...) do {} while(0) +#define devdbg(usbnet, fmt, arg...) \ + ({ if (0) printk(KERN_DEBUG "%s: " fmt "\n" , (usbnet)->net->name , \ + ## arg); 0; })#endif #define deverr(usbnet, fmt, arg...) \
From: David Brownell <[email protected]>
Date: Fri, 16 Jan 2009 23:19:44 -0800
Quoted 6 lines
Improve usbnet's devdbg to always type-check diagnostic arguments, like
dev_dbg(device.h). This makes no change to the resulting size of usbnet modules.Signed-off-by Steve Glendinning <[email protected]>Signed-off-by David Brownell <[email protected]>
Applied, thanks.
David Miller
maintainerFrom: David Miller <[email protected]>
Date: Mon, 19 Jan 2009 17:12:00 -0800 (PST)
Quoted 11 lines
From: David Brownell <[email protected]>
Date: Fri, 16 Jan 2009 23:19:44 -0800Quoted 6 lines
Improve usbnet's devdbg to always type-check diagnostic arguments, like
dev_dbg(device.h). This makes no change to the resulting size of usbnet modules.Signed-off-by Steve Glendinning <[email protected]>Signed-off-by David Brownell <[email protected]>Applied, thanks.
I have to back this out, it breaks the build:
drivers/net/wireless/rndis_wlan.c: In function 'rndis_translate_scan':drivers/net/wireless/rndis_wlan.c:1664: error: 'usbdev' undeclared (first use in this function) drivers/net/wireless/rndis_wlan.c:1664: error: (Each undeclared identifier is reported only once drivers/net/wireless/rndis_wlan.c:1664: error: for each function it appears in.) make[3]: *** [drivers/net/wireless/rndis_wlan.o] Error 1
make[3]: *** Waiting for unfinished jobs....
make[2]: *** [drivers/net/wireless] Error 2
make[2]: *** Waiting for unfinished jobs....
I bet there are other similar gremlins like this in the tree.
Quoted 26 lines · David Miller
From: David Miller <[email protected]>
Date: Mon, 19 Jan 2009 17:12:00 -0800 (PST)Quoted 11 lines
From: David Brownell <[email protected]>
Date: Fri, 16 Jan 2009 23:19:44 -0800Quoted 6 lines
Improve usbnet's devdbg to always type-check diagnostic arguments, like
dev_dbg(device.h). This makes no change to the resulting size of usbnet modules.Signed-off-by Steve Glendinning <[email protected]>Signed-off-by David Brownell <[email protected]>Applied, thanks.
I have to back this out, it breaks the build:
drivers/net/wireless/rndis_wlan.c: In function 'rndis_translate_scan':drivers/net/wireless/rndis_wlan.c:1664: error: 'usbdev' undeclared (first use in this function)drivers/net/wireless/rndis_wlan.c:1664: error: (Each undeclared identifier is reported only oncedrivers/net/wireless/rndis_wlan.c:1664: error: for each function it appears in.) make[3]: *** [drivers/net/wireless/rndis_wlan.o] Error 1
make[3]: *** Waiting for unfinished jobs....
make[2]: *** [drivers/net/wireless] Error 2
make[2]: *** Waiting for unfinished jobs....
Quoted 1 line
I bet there are other similar gremlins like this in the tree.
Maybe, but that's after all part of why we want this kind of patch merged ... after the first gremlines are fixed!
In this case I'll guess this is the root cause of the bug:
#ifdef DEBUG
struct usbnet *usbdev = netdev_priv(dev);#endif
CC'd someone more involved in rndis_wlan than me ...
- Dave
David Miller
maintainerFrom: David Brownell <[email protected]>
Date: Mon, 19 Jan 2009 18:04:08 -0800
Quoted 2 lines
Maybe, but that's after all part of why we want this kind of patch merged ... after the first gremlines are fixed!
When you submit a change like this that can have potential warning and build failure effects across the entire tree, the onus is on you to do an allmodconfig sanity check build or similar and add any necessary fixes to your change.
Quoted 8 lines · David Miller
Quoted 2 lines
Maybe, but that's after all part of why we want this kind of patch merged ... after the first gremlines are fixed!
When you submit a change like this that can have potential warning and build failure effects across the entire tree, the onus is on you to do an allmodconfig sanity check build or similar and add any necessary fixes to your change.
It's actually Steve's patch ... Steve?
David Brownell <[email protected]> wrote on 20/01/2009 05:11:48:
Quoted 12 lines · David Miller
Quoted 8 lines
Quoted 2 lines
Maybe, but that's after all part of why we want this kind of patch merged ... after the first gremlines are fixed!
When you submit a change like this that can have potential warning and build failure effects across the entire tree, the onus is on you to do an allmodconfig sanity check build or similar and add any necessary fixes to your change.
It's actually Steve's patch ... Steve?
Sorry, my bad - I tested compilation of usbnet drivers in drivers/net/usb but missed this wireless usbnet driver hiding in drivers/net/wireless. allmodconfig was what i was looking for, thanks David!
I've fixed this compilation breakage now, updated patchset to follow.
Steve
David Miller
maintainerFrom: [email protected]
Date: Tue, 20 Jan 2009 10:56:09 +0000
Quoted 2 lines
I've fixed this compilation breakage now, updated patchset to follow.
But you added a new warning for the non-DEBUG case during the intermediate steps between the patches.
Please respin this all into a single patch in order to make the transition cleanly, for all cases.