| Message ID | 20171025205855.30492-1-gael.portay@savoirfairelinux.com |
|---|---|
| State | Accepted |
| Headers | show |
| Series | Add mtd-utils dependency if cfi or ubivol are set | expand |
On Wednesday, 25 October 2017 16:59:00 UTC-4, Gaël PORTAY wrote: > This fixes the build issue below when CONFIG_CFI or CONFIG_UBIVOL are > set but CONFIG_MTD is not. > > | In file included from handlers/flash_handler.c:44:0: > | include/flash.h:26:24: fatal error: mtd/libmtd.h: No such file or directory > | #include <mtd/libmtd.h> > > Signed-off-by: Gaël PORTAY <gael.portay@savoirfairelinux.com> > > --- > > Hi all, > > As said into the commit message, this fixes a build issue when the > mtd-utils package is not built as dependency. > > The issue comes from the internal flash.h header which needs the header > mtd/libmtd.h from mtd-utils. > > This internal header is included by these 5 following .c source files: > > $ grep -r '#include.*flash.h' . > ./core/swupdate.c:#include "flash.h" > ./corelib/stream_interface.c:#include "flash.h" > ./corelib/mtd-interface.c:#include "flash.h" > ./handlers/ubivol_handler.c:#include "flash.h" > ./handlers/flash_handler.c:#include "flash.h" > > The file mtd-interface.c is compiled when CONFIG_MTD is set; this case > is handled by swupdate.inc (see python() function). > > lib-$(CONFIG_MTD) += mtd-interface.o > > Both files swupdate.c and stream_interface.c are also handled when > CONFIG_MTD is set. They include the flash.h header between the > following preprocessor block: > > #ifdef CONFIG_MTD > #include "flash.h" > #endif > > But, concerning flash_handler.c and ubivol_handler.c they are compiled > when their option is set CONFIG_CFI and CONFIG_UBIVOL. > > obj-$(CONFIG_CFI) += flash_handler.o > obj-$(CONFIG_UBIVOL) += ubivol_handler.o > > Thus the python() function from swupdate.inc needs to handle those new > two cases. > > Regards, > Gael > > recipes-support/swupdate/swupdate.inc | 7 +++++++ > 1 file changed, 7 insertions(+) > > diff --git a/recipes-support/swupdate/swupdate.inc b/recipes-support/swupdate/swupdate.inc > index 53a61ea..1431d15 100644 > --- a/recipes-support/swupdate/swupdate.inc > +++ b/recipes-support/swupdate/swupdate.inc > @@ -86,6 +86,13 @@ python () { > depends = d.getVar('DEPENDS', False) > d.setVar('DEPENDS', depends + ' mtd-utils') > > + if 'CONFIG_CFI=y\n' in features: > + depends = d.getVar('DEPENDS', False) > + d.setVar('DEPENDS', depends + ' mtd-utils') > + > + if 'CONFIG_UBIVOL=y\n' in features: > + depends = d.getVar('DEPENDS', False) > + d.setVar('DEPENDS', depends + ' mtd-utils') > } > > do_configure () { > -- > 2.14.2 This is for meta-swupdate. My appologies for the subject.
On 25/10/2017 22:58, Gaël PORTAY wrote: > This fixes the build issue below when CONFIG_CFI or CONFIG_UBIVOL are > set but CONFIG_MTD is not. > > | In file included from handlers/flash_handler.c:44:0: > | include/flash.h:26:24: fatal error: mtd/libmtd.h: No such file or directory > | #include <mtd/libmtd.h> > > Signed-off-by: Gaël PORTAY <gael.portay@savoirfairelinux.com> > > --- > > Hi all, > > As said into the commit message, this fixes a build issue when the > mtd-utils package is not built as dependency. > > The issue comes from the internal flash.h header which needs the header > mtd/libmtd.h from mtd-utils. > > This internal header is included by these 5 following .c source files: > > $ grep -r '#include.*flash.h' . > ./core/swupdate.c:#include "flash.h" > ./corelib/stream_interface.c:#include "flash.h" > ./corelib/mtd-interface.c:#include "flash.h" > ./handlers/ubivol_handler.c:#include "flash.h" > ./handlers/flash_handler.c:#include "flash.h" > > The file mtd-interface.c is compiled when CONFIG_MTD is set; this case > is handled by swupdate.inc (see python() function). > > lib-$(CONFIG_MTD) += mtd-interface.o > > Both files swupdate.c and stream_interface.c are also handled when > CONFIG_MTD is set. They include the flash.h header between the > following preprocessor block: > > #ifdef CONFIG_MTD > #include "flash.h" > #endif > > But, concerning flash_handler.c and ubivol_handler.c they are compiled > when their option is set CONFIG_CFI and CONFIG_UBIVOL. > > obj-$(CONFIG_CFI) += flash_handler.o > obj-$(CONFIG_UBIVOL) += ubivol_handler.o > > Thus the python() function from swupdate.inc needs to handle those new > two cases. > > Regards, > Gael > > recipes-support/swupdate/swupdate.inc | 7 +++++++ > 1 file changed, 7 insertions(+) > > diff --git a/recipes-support/swupdate/swupdate.inc b/recipes-support/swupdate/swupdate.inc > index 53a61ea..1431d15 100644 > --- a/recipes-support/swupdate/swupdate.inc > +++ b/recipes-support/swupdate/swupdate.inc > @@ -86,6 +86,13 @@ python () { > depends = d.getVar('DEPENDS', False) > d.setVar('DEPENDS', depends + ' mtd-utils') > > + if 'CONFIG_CFI=y\n' in features: > + depends = d.getVar('DEPENDS', False) > + d.setVar('DEPENDS', depends + ' mtd-utils') > + > + if 'CONFIG_UBIVOL=y\n' in features: > + depends = d.getVar('DEPENDS', False) > + d.setVar('DEPENDS', depends + ' mtd-utils') > } > > do_configure () { > Applied to -master, -pyro and the new -rocko, thanks ! Best regards, Stefano Babic
Hi Stefano, On Mon, Oct 30, 2017 at 02:58:47PM +0100, Stefano Babic wrote: > (...) > > Applied to -master, -pyro and the new -rocko, thanks ! > Thanks you! Since I am using -morty, could I ask you to apply it to -morty as well? I really like to use upstream layers with no additional commits :) Thank you in advance. King Regards, Gael
On 30/10/2017 18:38, Gaël PORTAY wrote: > Hi Stefano, > > On Mon, Oct 30, 2017 at 02:58:47PM +0100, Stefano Babic wrote: >> (...) >> >> Applied to -master, -pyro and the new -rocko, thanks ! >> > > Thanks you! > > Since I am using -morty, could I ask you to apply it to -morty as well? > > I really like to use upstream layers with no additional commits :) > > Thank you in advance. > Done. Best regards, Stefano Babic
On Mon, Oct 30, 2017 at 06:44:36PM +0100, Stefano Babic wrote: > On 30/10/2017 18:38, Gaël PORTAY wrote: > > Hi Stefano, > > > > On Mon, Oct 30, 2017 at 02:58:47PM +0100, Stefano Babic wrote: > >> (...) > >> > >> Applied to -master, -pyro and the new -rocko, thanks ! > >> > > > > Thanks you! > > > > Since I am using -morty, could I ask you to apply it to -morty as well? > > > > I really like to use upstream layers with no additional commits :) > > > > Thank you in advance. > > > Thank you, King Regards, Gael
diff --git a/recipes-support/swupdate/swupdate.inc b/recipes-support/swupdate/swupdate.inc index 53a61ea..1431d15 100644 --- a/recipes-support/swupdate/swupdate.inc +++ b/recipes-support/swupdate/swupdate.inc @@ -86,6 +86,13 @@ python () { depends = d.getVar('DEPENDS', False) d.setVar('DEPENDS', depends + ' mtd-utils') + if 'CONFIG_CFI=y\n' in features: + depends = d.getVar('DEPENDS', False) + d.setVar('DEPENDS', depends + ' mtd-utils') + + if 'CONFIG_UBIVOL=y\n' in features: + depends = d.getVar('DEPENDS', False) + d.setVar('DEPENDS', depends + ' mtd-utils') } do_configure () {
This fixes the build issue below when CONFIG_CFI or CONFIG_UBIVOL are set but CONFIG_MTD is not. | In file included from handlers/flash_handler.c:44:0: | include/flash.h:26:24: fatal error: mtd/libmtd.h: No such file or directory | #include <mtd/libmtd.h> Signed-off-by: Gaël PORTAY <gael.portay@savoirfairelinux.com> --- Hi all, As said into the commit message, this fixes a build issue when the mtd-utils package is not built as dependency. The issue comes from the internal flash.h header which needs the header mtd/libmtd.h from mtd-utils. This internal header is included by these 5 following .c source files: $ grep -r '#include.*flash.h' . ./core/swupdate.c:#include "flash.h" ./corelib/stream_interface.c:#include "flash.h" ./corelib/mtd-interface.c:#include "flash.h" ./handlers/ubivol_handler.c:#include "flash.h" ./handlers/flash_handler.c:#include "flash.h" The file mtd-interface.c is compiled when CONFIG_MTD is set; this case is handled by swupdate.inc (see python() function). lib-$(CONFIG_MTD) += mtd-interface.o Both files swupdate.c and stream_interface.c are also handled when CONFIG_MTD is set. They include the flash.h header between the following preprocessor block: #ifdef CONFIG_MTD #include "flash.h" #endif But, concerning flash_handler.c and ubivol_handler.c they are compiled when their option is set CONFIG_CFI and CONFIG_UBIVOL. obj-$(CONFIG_CFI) += flash_handler.o obj-$(CONFIG_UBIVOL) += ubivol_handler.o Thus the python() function from swupdate.inc needs to handle those new two cases. Regards, Gael recipes-support/swupdate/swupdate.inc | 7 +++++++ 1 file changed, 7 insertions(+)