Hi, On my x64 debian build box, after make distclean hg purge --all ./configure make -j8 deb the build eventually fails with this:> FV Space Information > MAINFV [74%Full] 5242880 total, 3888168 used, 1354712 free > SECFV [86%Full] 81920 total, 70792 used, 11128 free > FVMAIN_COMPACT [74%Full] 966656 total, 719808 used, 246848 free > DXEFV [99%Full] 3670016 total, 3669544 used, 472 free > make[7]: Leaving directory > `/local/scratch/tdeegan/xen-unstable.hg/tools/firmware/ovmf-remote/Build/OvmfX64/DEBUG_GCC46'' > > - Done - > Build end time: 12:46:06, Mar.22 2012 > Build total time: 00:00:47 > > cp Build/OvmfX64/DEBUG_GCC44/FV/OVMF.fd ovmf.bin > cp: cannot stat `Build/OvmfX64/DEBUG_GCC44/FV/OVMF.fd'': No such file or directory > make[6]: *** [ovmf.bin] Error 1 > make[6]: Leaving directory `/local/scratch/tdeegan/xen-unstable.hg/tools/firmware/ovmf-remote''Using make -j1 doesn''t unstick it - I haven''t yet tried a -j1 build from clean. I can''t see any errors earlier in the build process. Is there a ./configure rune to disable OVMF? Cheers, Tim.
On 22/03/12 12:59, Tim Deegan wrote:> Hi, > > On my x64 debian build box, after > make distclean > hg purge --all > ./configure > make -j8 deb > > the build eventually fails with this: > > >> FV Space Information >> MAINFV [74%Full] 5242880 total, 3888168 used, 1354712 free >> SECFV [86%Full] 81920 total, 70792 used, 11128 free >> FVMAIN_COMPACT [74%Full] 966656 total, 719808 used, 246848 free >> DXEFV [99%Full] 3670016 total, 3669544 used, 472 free >> make[7]: Leaving directory >> `/local/scratch/tdeegan/xen-unstable.hg/tools/firmware/ovmf-remote/Build/OvmfX64/DEBUG_GCC46'' >> >> - Done - >> Build end time: 12:46:06, Mar.22 2012 >> Build total time: 00:00:47 >> >> cp Build/OvmfX64/DEBUG_GCC44/FV/OVMF.fd ovmf.bin >> cp: cannot stat `Build/OvmfX64/DEBUG_GCC44/FV/OVMF.fd'': No such file or directory >> make[6]: *** [ovmf.bin] Error 1 >> make[6]: Leaving directory `/local/scratch/tdeegan/xen-unstable.hg/tools/firmware/ovmf-remote'' >> > Using make -j1 doesn''t unstick it - I haven''t yet tried a -j1 build from > clean. I can''t see any errors earlier in the build process. > > Is there a ./configure rune to disable OVMF? > >Actually CONFIG_OVMF should be disabled by default, can you please check how is in your case? Also, did you apply the OVMF v4 patch for enabling the CONFIG_OVMF on? Attilio
At 13:07 +0000 on 22 Mar (1332421679), Attilio Rao wrote:> Actually CONFIG_OVMF should be disabled by default, can you please check > how is in your case?whitby:xen-unstable.hg$ grep CONFIG_OVMF Config.mk CONFIG_OVMF ?= $(CONFIG_Linux) changeset: 25077:6bf50858c3c5 user: Attilio Rao <attilio.rao@citrix.com> date: Thu Mar 22 10:26:03 2012 +0000 summary: Fetch the OVMF repository from specific git mirror and enable it> Also, did you apply the OVMF v4 patch for enabling the CONFIG_OVMF on?Nope, AFAICT I have no local patches. Tim.
On 22/03/2012 13:15, "Tim Deegan" <tim@xen.org> wrote:> At 13:07 +0000 on 22 Mar (1332421679), Attilio Rao wrote: >> Actually CONFIG_OVMF should be disabled by default, can you please check >> how is in your case? > > whitby:xen-unstable.hg$ grep CONFIG_OVMF Config.mk > CONFIG_OVMF ?= $(CONFIG_Linux) > > changeset: 25077:6bf50858c3c5 > user: Attilio Rao <attilio.rao@citrix.com> > date: Thu Mar 22 10:26:03 2012 +0000 > summary: Fetch the OVMF repository from specific git mirror and enable > it > >> Also, did you apply the OVMF v4 patch for enabling the CONFIG_OVMF on? > > Nope, AFAICT I have no local patches.I applied the OVMF-enable patch this morning, hence Tim''s problem. I might have applied v3 rather than v4 though? Please check. -- Keir> > Tim. > > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.xen.org > http://lists.xen.org/xen-devel
On Thu, 2012-03-22 at 13:58 +0000, Keir Fraser wrote:> On 22/03/2012 13:15, "Tim Deegan" <tim@xen.org> wrote: > > > At 13:07 +0000 on 22 Mar (1332421679), Attilio Rao wrote: > >> Actually CONFIG_OVMF should be disabled by default, can you please check > >> how is in your case? > > > > whitby:xen-unstable.hg$ grep CONFIG_OVMF Config.mk > > CONFIG_OVMF ?= $(CONFIG_Linux) > > > > changeset: 25077:6bf50858c3c5 > > user: Attilio Rao <attilio.rao@citrix.com> > > date: Thu Mar 22 10:26:03 2012 +0000 > > summary: Fetch the OVMF repository from specific git mirror and enable > > it > > > >> Also, did you apply the OVMF v4 patch for enabling the CONFIG_OVMF on? > > > > Nope, AFAICT I have no local patches. > > I applied the OVMF-enable patch this morning, hence Tim''s problem. I might > have applied v3 rather than v4 though? Please check.I think that was supposed to be an RFC pending a build fix getting committed to OVMF upstream.> > -- Keir > > > > > Tim. > > > > _______________________________________________ > > Xen-devel mailing list > > Xen-devel@lists.xen.org > > http://lists.xen.org/xen-devel > > > > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.xen.org > http://lists.xen.org/xen-devel
On 22/03/12 14:08, Ian Campbell wrote:> On Thu, 2012-03-22 at 13:58 +0000, Keir Fraser wrote: > >> On 22/03/2012 13:15, "Tim Deegan"<tim@xen.org> wrote: >> >> >>> At 13:07 +0000 on 22 Mar (1332421679), Attilio Rao wrote: >>> >>>> Actually CONFIG_OVMF should be disabled by default, can you please check >>>> how is in your case? >>>> >>> whitby:xen-unstable.hg$ grep CONFIG_OVMF Config.mk >>> CONFIG_OVMF ?= $(CONFIG_Linux) >>> >>> changeset: 25077:6bf50858c3c5 >>> user: Attilio Rao<attilio.rao@citrix.com> >>> date: Thu Mar 22 10:26:03 2012 +0000 >>> summary: Fetch the OVMF repository from specific git mirror and enable >>> it >>> >>> >>>> Also, did you apply the OVMF v4 patch for enabling the CONFIG_OVMF on? >>>> >>> Nope, AFAICT I have no local patches. >>> >> I applied the OVMF-enable patch this morning, hence Tim''s problem. I might >> have applied v3 rather than v4 though? Please check. >> > I think that was supposed to be an RFC pending a build fix getting > committed to OVMF upstream. > >Yes, we need v4 which also contains explicit showstopping informations based on a OVMF patch pending. Thus, likely we should revert v3 inclusion, wait another couple of days for Tianocore people to reply, eventually push the tianocore patch in our mirror and in the end apply v4. Attilio
At 12:59 +0000 on 22 Mar (1332421197), Tim Deegan wrote:> > cp Build/OvmfX64/DEBUG_GCC44/FV/OVMF.fd ovmf.bin > > cp: cannot stat `Build/OvmfX64/DEBUG_GCC44/FV/OVMF.fd'': No such file or directory > > make[6]: *** [ovmf.bin] Error 1 > > make[6]: Leaving directory `/local/scratch/tdeegan/xen-unstable.hg/tools/firmware/ovmf-remote''Also, ''make clean'' triggers the following code in tools/firmware/seabios-dir-remote/tools/test-gcc.sh : if [ $? -ne 0 ]; then echo " Working around no -fwhole-program" > /dev/fd/2 echo 2 exit 0 fi which truncates std err and mangles my build log. FTW! # HG changeset patch # User Tim Deegan <tim@xen.org> # Date 1332428104 0 # Node ID 6e34d38fda90a3a670fcd231e01d172dee3c3229 # Parent d5ccb2d1dbd1dce76f9d427073306941d48c3095 Disable OVMF build - it''s not baked yet. Signed-off-by: Tim Deegan <tim@xen.org> diff -r d5ccb2d1dbd1 -r 6e34d38fda90 Config.mk --- a/Config.mk Thu Mar 22 12:20:13 2012 +0000 +++ b/Config.mk Thu Mar 22 14:55:04 2012 +0000 @@ -207,7 +207,7 @@ SEABIOS_UPSTREAM_TAG ?= rel-1.6.3.2 ETHERBOOT_NICS ?= rtl8139 8086100e -CONFIG_OVMF ?= $(CONFIG_Linux) +CONFIG_OVMF ?= n CONFIG_ROMBIOS ?= y CONFIG_SEABIOS ?= y
On Thu, 2012-03-22 at 14:55 +0000, Tim Deegan wrote:> At 12:59 +0000 on 22 Mar (1332421197), Tim Deegan wrote: > > > cp Build/OvmfX64/DEBUG_GCC44/FV/OVMF.fd ovmf.bin > > > cp: cannot stat `Build/OvmfX64/DEBUG_GCC44/FV/OVMF.fd'': No such file or directory > > > make[6]: *** [ovmf.bin] Error 1 > > > make[6]: Leaving directory `/local/scratch/tdeegan/xen-unstable.hg/tools/firmware/ovmf-remote'' > > Also, ''make clean'' triggers the following code in > tools/firmware/seabios-dir-remote/tools/test-gcc.sh : > > if [ $? -ne 0 ]; then > echo " Working around no -fwhole-program" > /dev/fd/2 > echo 2 > exit 0 > fiI suppose making that ">>" fixes it for you? Likewise "1>&2"? If you can confirm that they work I''ll send a patch to seabios.> which truncates std err and mangles my build log. FTW!Quite.> > # HG changeset patch > # User Tim Deegan <tim@xen.org> > # Date 1332428104 0 > # Node ID 6e34d38fda90a3a670fcd231e01d172dee3c3229 > # Parent d5ccb2d1dbd1dce76f9d427073306941d48c3095 > Disable OVMF build - it''s not baked yet. > > Signed-off-by: Tim Deegan <tim@xen.org> > > diff -r d5ccb2d1dbd1 -r 6e34d38fda90 Config.mk > --- a/Config.mk Thu Mar 22 12:20:13 2012 +0000 > +++ b/Config.mk Thu Mar 22 14:55:04 2012 +0000 > @@ -207,7 +207,7 @@ SEABIOS_UPSTREAM_TAG ?= rel-1.6.3.2 > > ETHERBOOT_NICS ?= rtl8139 8086100e > > -CONFIG_OVMF ?= $(CONFIG_Linux) > +CONFIG_OVMF ?= n > CONFIG_ROMBIOS ?= y > CONFIG_SEABIOS ?= y > > > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.xen.org > http://lists.xen.org/xen-devel
At 15:04 +0000 on 22 Mar (1332428686), Ian Campbell wrote:> On Thu, 2012-03-22 at 14:55 +0000, Tim Deegan wrote: > > At 12:59 +0000 on 22 Mar (1332421197), Tim Deegan wrote: > > > > cp Build/OvmfX64/DEBUG_GCC44/FV/OVMF.fd ovmf.bin > > > > cp: cannot stat `Build/OvmfX64/DEBUG_GCC44/FV/OVMF.fd'': No such file or directory > > > > make[6]: *** [ovmf.bin] Error 1 > > > > make[6]: Leaving directory `/local/scratch/tdeegan/xen-unstable.hg/tools/firmware/ovmf-remote'' > > > > Also, ''make clean'' triggers the following code in > > tools/firmware/seabios-dir-remote/tools/test-gcc.sh : > > > > if [ $? -ne 0 ]; then > > echo " Working around no -fwhole-program" > /dev/fd/2 > > echo 2 > > exit 0 > > fi > > I suppose making that ">>" fixes it for you? > > Likewise "1>&2"?Yeah, >&2 works fine (in a bunch of places in that file) Tim.
On 22/03/12 14:55, Tim Deegan wrote:> At 12:59 +0000 on 22 Mar (1332421197), Tim Deegan wrote: > >>> cp Build/OvmfX64/DEBUG_GCC44/FV/OVMF.fd ovmf.bin >>> cp: cannot stat `Build/OvmfX64/DEBUG_GCC44/FV/OVMF.fd'': No such file or directory >>> make[6]: *** [ovmf.bin] Error 1 >>> make[6]: Leaving directory `/local/scratch/tdeegan/xen-unstable.hg/tools/firmware/ovmf-remote'' >>> > Also, ''make clean'' triggers the following code in > tools/firmware/seabios-dir-remote/tools/test-gcc.sh : > > if [ $? -ne 0 ]; then > echo " Working around no -fwhole-program"> /dev/fd/2 > echo 2 > exit 0 > fi > > which truncates std err and mangles my build log. FTW! > > # HG changeset patch > # User Tim Deegan<tim@xen.org> > # Date 1332428104 0 > # Node ID 6e34d38fda90a3a670fcd231e01d172dee3c3229 > # Parent d5ccb2d1dbd1dce76f9d427073306941d48c3095 > Disable OVMF build - it''s not baked yet. > > Signed-off-by: Tim Deegan<tim@xen.org> >Ack''ed by me. Attilio
On 22/03/12 13:58, Keir Fraser wrote:> On 22/03/2012 13:15, "Tim Deegan"<tim@xen.org> wrote: > > >> At 13:07 +0000 on 22 Mar (1332421679), Attilio Rao wrote: >> >>> Actually CONFIG_OVMF should be disabled by default, can you please check >>> how is in your case? >>> >> whitby:xen-unstable.hg$ grep CONFIG_OVMF Config.mk >> CONFIG_OVMF ?= $(CONFIG_Linux) >> >> changeset: 25077:6bf50858c3c5 >> user: Attilio Rao<attilio.rao@citrix.com> >> date: Thu Mar 22 10:26:03 2012 +0000 >> summary: Fetch the OVMF repository from specific git mirror and enable >> it >> >> >>> Also, did you apply the OVMF v4 patch for enabling the CONFIG_OVMF on? >>> >> Nope, AFAICT I have no local patches. >> > I applied the OVMF-enable patch this morning, hence Tim''s problem. I might > have applied v3 rather than v4 though? Please check. > >I actually think you have applied the right patch, there is just a situation in OVMF binary building that we need to cope with, my bad for not having better checked. For the moment, please disable OVMF by default building, I''ll send you more patches and when the upstream TIANOCORE will have fixes in I will send an explicit patch to enable it. Thanks, Attilio
On Thu, 2012-03-22 at 15:16 +0000, Tim Deegan wrote:> At 15:04 +0000 on 22 Mar (1332428686), Ian Campbell wrote: > > On Thu, 2012-03-22 at 14:55 +0000, Tim Deegan wrote: > > > At 12:59 +0000 on 22 Mar (1332421197), Tim Deegan wrote: > > > > > cp Build/OvmfX64/DEBUG_GCC44/FV/OVMF.fd ovmf.bin > > > > > cp: cannot stat `Build/OvmfX64/DEBUG_GCC44/FV/OVMF.fd'': No such file or directory > > > > > make[6]: *** [ovmf.bin] Error 1 > > > > > make[6]: Leaving directory `/local/scratch/tdeegan/xen-unstable.hg/tools/firmware/ovmf-remote'' > > > > > > Also, ''make clean'' triggers the following code in > > > tools/firmware/seabios-dir-remote/tools/test-gcc.sh : > > > > > > if [ $? -ne 0 ]; then > > > echo " Working around no -fwhole-program" > /dev/fd/2 > > > echo 2 > > > exit 0 > > > fi > > > > I suppose making that ">>" fixes it for you? > > > > Likewise "1>&2"? > > Yeah, >&2 works fine (in a bunch of places in that file)The fix has been accepted into seabios mainline. The "reported-by..." didn''t get turned into a CC by git send-email like I thought it would. Ian.
On Thu, Mar 22, 2012 at 05:59, Tim Deegan <tim@xen.org> wrote:> Hi, > > On my x64 debian build box, after > make distclean > hg purge --all > ./configure > make -j8 deb > > the build eventually fails with this: > >> FV Space Information >> MAINFV [74%Full] 5242880 total, 3888168 used, 1354712 free >> SECFV [86%Full] 81920 total, 70792 used, 11128 free >> FVMAIN_COMPACT [74%Full] 966656 total, 719808 used, 246848 free >> DXEFV [99%Full] 3670016 total, 3669544 used, 472 free >> make[7]: Leaving directory >> `/local/scratch/tdeegan/xen-unstable.hg/tools/firmware/ovmf-remote/Build/OvmfX64/DEBUG_GCC46'' >> >> - Done - >> Build end time: 12:46:06, Mar.22 2012 >> Build total time: 00:00:47 >> >> cp Build/OvmfX64/DEBUG_GCC44/FV/OVMF.fd ovmf.bin >> cp: cannot stat `Build/OvmfX64/DEBUG_GCC44/FV/OVMF.fd'': No such file or directoryLooks like GCC46 was used during the build here, so DEBUG_GCC46 is the output path, not DEBUG_GCC44. -Jordan