* [PATCH] media_build: fix module_*_driver redefined warnings @ 2012-03-20 14:10 Gianluca Gennari 2012-04-09 12:34 ` Mauro Carvalho Chehab 0 siblings, 1 reply; 4+ messages in thread From: Gianluca Gennari @ 2012-03-20 14:10 UTC (permalink / raw) To: linux-media, mchehab; +Cc: Gianluca Gennari The conditions "#ifndef module_usb_driver" and "#ifndef module_platform_driver" are always true, as the header files where this macros are defined are not included in compat.h (linux/usb.h and linux/platform_devices.h). This produces a lot of warnings like "module_usb_driver redefined" or "module_platform_driver redefined" with kernels 3.2 and 3.3. But including the header files in compat.h produces other "redefined" warnings, so let's check the kernel version instead. module_usb_driver was first introduced in kernel 3.3, while module_platform_driver was introduced in kernel 3.2. Tested with kernel 3.3, 3.2 and 3.0. Signed-off-by: Gianluca Gennari <gennarone@gmail.com> --- v4l/compat.h | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) diff --git a/v4l/compat.h b/v4l/compat.h index 62710c9..ab0f2e7 100644 --- a/v4l/compat.h +++ b/v4l/compat.h @@ -864,7 +864,7 @@ static inline int snd_ctl_enum_info(struct snd_ctl_elem_info *info, unsigned int #endif #endif /*pr_debug_ratelimited */ -#ifndef module_usb_driver +#if LINUX_VERSION_CODE < KERNEL_VERSION(3, 3, 0) #define module_usb_driver(drv) \ static int __init usb_mod_init(void) \ { \ @@ -878,7 +878,7 @@ module_init(usb_mod_init); \ module_exit(usb_mod_exit); #endif /* module_usb_driver */ -#ifndef module_platform_driver +#if LINUX_VERSION_CODE < KERNEL_VERSION(3, 2, 0) #define module_platform_driver(drv) \ static int __init plat_mod_init(void) \ { \ -- 1.7.5.4 ^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH] media_build: fix module_*_driver redefined warnings 2012-03-20 14:10 [PATCH] media_build: fix module_*_driver redefined warnings Gianluca Gennari @ 2012-04-09 12:34 ` Mauro Carvalho Chehab 2012-04-09 14:15 ` Gianluca Gennari 0 siblings, 1 reply; 4+ messages in thread From: Mauro Carvalho Chehab @ 2012-04-09 12:34 UTC (permalink / raw) To: Gianluca Gennari; +Cc: linux-media Hi Gianluca, Em 20-03-2012 11:10, Gianluca Gennari escreveu: > The conditions "#ifndef module_usb_driver" and "#ifndef module_platform_driver" > are always true, as the header files where this macros are defined are not > included in compat.h (linux/usb.h and linux/platform_devices.h). > > This produces a lot of warnings like "module_usb_driver redefined" or > "module_platform_driver redefined" with kernels 3.2 and 3.3. > > But including the header files in compat.h produces other "redefined" warnings, > so let's check the kernel version instead. > > module_usb_driver was first introduced in kernel 3.3, > while module_platform_driver was introduced in kernel 3.2. > > Tested with kernel 3.3, 3.2 and 3.0. > > Signed-off-by: Gianluca Gennari <gennarone@gmail.com> > --- > v4l/compat.h | 4 ++-- > 1 files changed, 2 insertions(+), 2 deletions(-) > > diff --git a/v4l/compat.h b/v4l/compat.h > index 62710c9..ab0f2e7 100644 > --- a/v4l/compat.h > +++ b/v4l/compat.h > @@ -864,7 +864,7 @@ static inline int snd_ctl_enum_info(struct snd_ctl_elem_info *info, unsigned int > #endif > #endif /*pr_debug_ratelimited */ > > -#ifndef module_usb_driver > +#if LINUX_VERSION_CODE < KERNEL_VERSION(3, 3, 0) Please avoid adding more tests for an specific Kernel version here. There are two issues with checks like that: 1) this may break on some kernel-fix release that might backport the function. This is not very common, but there was some cases like that, in the USB subsystem; 2) this generally breaks compilation, after some time, if someone tries to compile it against a distribution-patched kernel, as the new code may be backported there. That's said, if just doing an "#ifdef module_usb_driver" doesn't work because this is not a macro, you can add a simple check at this script: v4l/scripts/make_config_compat.pl like this one: check_file_for_func("include/linux/delay.h", "usleep_range", "NEED_USLEEP_RANGE"); This function will seek for "usleep_range" at the delay.h header. If not found, it will add a #define NEED_USLEEP_RANGE at v4l/config-compat.h, that can be checked inside compat.h: #ifdef NEED_USLEEP_RANGE #define usleep_range(min, max) msleep(min/1000) #endif You can use the same kind of logic for module_usb_driver. Regards, Mauro > #define module_usb_driver(drv) \ > static int __init usb_mod_init(void) \ > { \ > @@ -878,7 +878,7 @@ module_init(usb_mod_init); \ > module_exit(usb_mod_exit); > #endif /* module_usb_driver */ > > -#ifndef module_platform_driver > +#if LINUX_VERSION_CODE < KERNEL_VERSION(3, 2, 0) > #define module_platform_driver(drv) \ > static int __init plat_mod_init(void) \ > { \ ^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] media_build: fix module_*_driver redefined warnings 2012-04-09 12:34 ` Mauro Carvalho Chehab @ 2012-04-09 14:15 ` Gianluca Gennari 2012-04-09 16:31 ` Mauro Carvalho Chehab 0 siblings, 1 reply; 4+ messages in thread From: Gianluca Gennari @ 2012-04-09 14:15 UTC (permalink / raw) To: Mauro Carvalho Chehab; +Cc: linux-media Il 09/04/2012 14:34, Mauro Carvalho Chehab ha scritto: > Hi Gianluca, > > Em 20-03-2012 11:10, Gianluca Gennari escreveu: > > Please avoid adding more tests for an specific Kernel version here. There are > two issues with checks like that: > > 1) this may break on some kernel-fix release that might backport the function. > This is not very common, but there was some cases like that, in the USB subsystem; > > 2) this generally breaks compilation, after some time, if someone tries > to compile it against a distribution-patched kernel, as the new code may be > backported there. > > That's said, if just doing an "#ifdef module_usb_driver" doesn't work because this > is not a macro, you can add a simple check at this script: > v4l/scripts/make_config_compat.pl > > like this one: > > check_file_for_func("include/linux/delay.h", "usleep_range", "NEED_USLEEP_RANGE"); > > This function will seek for "usleep_range" at the delay.h header. If not found, it will > add a #define NEED_USLEEP_RANGE at v4l/config-compat.h, that can be checked inside compat.h: > > #ifdef NEED_USLEEP_RANGE > #define usleep_range(min, max) msleep(min/1000) > #endif > > You can use the same kind of logic for module_usb_driver. > > Regards, > Mauro Hi Mauro, thanks for the explanation but Hans Verkuil already solved the issue using the check_file_for_func method: http://git.linuxtv.org/media_build.git/commit/2492bf186743a925db98694911649fa0e94003f5 Of course, I agree this is a much better solution. Regards, Gianluca ^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] media_build: fix module_*_driver redefined warnings 2012-04-09 14:15 ` Gianluca Gennari @ 2012-04-09 16:31 ` Mauro Carvalho Chehab 0 siblings, 0 replies; 4+ messages in thread From: Mauro Carvalho Chehab @ 2012-04-09 16:31 UTC (permalink / raw) To: gennarone; +Cc: linux-media Em 09-04-2012 11:15, Gianluca Gennari escreveu: > > Il 09/04/2012 14:34, Mauro Carvalho Chehab ha scritto: >> Hi Gianluca, >> >> Em 20-03-2012 11:10, Gianluca Gennari escreveu: >> >> Please avoid adding more tests for an specific Kernel version here. There are >> two issues with checks like that: >> >> 1) this may break on some kernel-fix release that might backport the function. >> This is not very common, but there was some cases like that, in the USB subsystem; >> >> 2) this generally breaks compilation, after some time, if someone tries >> to compile it against a distribution-patched kernel, as the new code may be >> backported there. >> >> That's said, if just doing an "#ifdef module_usb_driver" doesn't work because this >> is not a macro, you can add a simple check at this script: >> v4l/scripts/make_config_compat.pl >> >> like this one: >> >> check_file_for_func("include/linux/delay.h", "usleep_range", "NEED_USLEEP_RANGE"); >> >> This function will seek for "usleep_range" at the delay.h header. If not found, it will >> add a #define NEED_USLEEP_RANGE at v4l/config-compat.h, that can be checked inside compat.h: >> >> #ifdef NEED_USLEEP_RANGE >> #define usleep_range(min, max) msleep(min/1000) >> #endif >> >> You can use the same kind of logic for module_usb_driver. >> >> Regards, >> Mauro > > Hi Mauro, > thanks for the explanation but Hans Verkuil already solved the issue > using the check_file_for_func method: > > http://git.linuxtv.org/media_build.git/commit/2492bf186743a925db98694911649fa0e94003f5 Yeah, I noticed when I've updated it to apply a patch for Debian handling. Anyway, I hope that the explanation is useful for you and others, as this way means less headache on maintaining backward compatibility. > Of course, I agree this is a much better solution. > > Regards, > Gianluca Regards, Mauro ^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2012-04-09 16:31 UTC | newest] Thread overview: 4+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2012-03-20 14:10 [PATCH] media_build: fix module_*_driver redefined warnings Gianluca Gennari 2012-04-09 12:34 ` Mauro Carvalho Chehab 2012-04-09 14:15 ` Gianluca Gennari 2012-04-09 16:31 ` Mauro Carvalho Chehab
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).