Bugzilla Bug 148072

Due to massive internal code changes and kernel-2.6 dependency, your keywords
have been dropped. Please test and re-keyword :)

anyone sees this with an up to date tree, too?



markus@powerbook ~/gentoo-x86/app-misc/lcdproc $ repoman scan

Setting paths:
PORTDIR = "/home/markus/gentoo-x86"
PORTDIR_OVERLAY = ""

RepoMan scours the neighborhood...

  DEPEND.bad                     4
   app-misc/lcdproc/lcdproc-0.5.0.ebuild: ~x86(default-linux/x86/2006.1)
['media-libs/libirman']
   app-misc/lcdproc/lcdproc-0.5.0.ebuild: ~x86(default-linux/x86/no-nptl)
['media-libs/libirman']
   app-misc/lcdproc/lcdproc-0.5.0.ebuild:
~x86(default-linux/x86/2006.1/desktop) ['media-libs/libirman']
   app-misc/lcdproc/lcdproc-0.5.0.ebuild: ~x86(hardened/x86/2.6)
['media-libs/libirman']
  RDEPEND.bad                    4
   app-misc/lcdproc/lcdproc-0.5.0.ebuild: ~x86(default-linux/x86/2006.1)
['media-libs/libirman']
   app-misc/lcdproc/lcdproc-0.5.0.ebuild: ~x86(default-linux/x86/no-nptl)
['media-libs/libirman']
   app-misc/lcdproc/lcdproc-0.5.0.ebuild:
~x86(default-linux/x86/2006.1/desktop) ['media-libs/libirman']
   app-misc/lcdproc/lcdproc-0.5.0.ebuild: ~x86(hardened/x86/2.6)
['media-libs/libirman']
  digest.assumed                 2
   digest-lcdproc-0.4.5::lcdproc-0.4.5.tar.bz2
   digest-lcdproc-0.5.0::lcdproc-0.5.0.tar.gz

markus@powerbook ~/gentoo-x86/app-misc/lcdproc $

ah.. libirman is masked:

# Alec Warner <antarus@gentoo.org> (19 Sep 2006)
# masked media-libs/libirman for treecleaners, bug(s) 89663
# Removal Oct 19th
media-libs/libirman

stupid bugs..

After discussion in bug #89663, lcdproc is unmasked again and has the USE flag
'irman' in base/use.mask, so the problem found by Markus in comment #1 should
be gone.

added ~ppc64

Emerged till ebuild on my ~amd64 system using ACCEPT_KEYWORDS="~x86" and it
worked just fine. So you could probably add ~amd64.

amd64,sparc,ppc: something holding this back?

yeah, time.

lcdproc-0.5.0-r1 does not compile with linux-headers-2.6.11-r2 (latest stable
on amd64), 2.6.17-r1 (latest ~amd64) works.
Here's the error:
if x86_64-pc-linux-gnu-gcc -DHAVE_CONFIG_H -I. -I. -I../..  -I../..  -fPIC
-Wall -march=k8 -O2 -pipe -DSEAMLESS_HBARS -DCFONTZ633_WRITE_DELAY=50
-Wno-unused-function -MT icp_a106.o -MD -MP -MF ".deps/icp_a106.Tpo" \
          -c -o icp_a106.o `test -f 'icp_a106.c' || echo './'`icp_a106.c; \
        then mv -f ".deps/icp_a106.Tpo" ".deps/icp_a106.Po"; \
        else rm -f ".deps/icp_a106.Tpo"; exit 1; \
        fi
In file included from /usr/include/linux/i2c-dev.h:28,
                 from hd44780-i2c.c:62:
/usr/include/linux/i2c.h:66: error: array type has incomplete element type
make[3]: *** [hd44780-i2c.o] Error 1
make[3]: *** Waiting for unfinished jobs....
make[3]: Leaving directory
`/var/tmp/portage/lcdproc-0.5.0-r1/work/lcdproc-0.5.0/server/drivers'
make[2]: *** [all-recursive] Error 1
make[2]: Leaving directory
`/var/tmp/portage/lcdproc-0.5.0-r1/work/lcdproc-0.5.0/server'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory
`/var/tmp/portage/lcdproc-0.5.0-r1/work/lcdproc-0.5.0'
make: *** [all] Error 2

amd64 was already testing, nothing to see here

Marked 0.5.0-r2 ~ppc, but it only compiles in my ~ppc chroot, ppc stable fails
with the error above.  0.5.1 fails in both chroots.

At least one driver is using asm (serialVFD includes server/drivers/port.h,
just look at line 293 and more). Trying lcdproc-0.5.1 btw.
Is driver exclusion per-arch possible?

if sparc-unknown-linux-gnu-gcc -DHAVE_CONFIG_H -I. -I. -I../..  -I../..  -fPIC
-Wall -mcpu=ultrasparc -mtune=ultrasparc -mvis -Wa,-Av8plusa -O2 -pipe
-Wno-unused-function -MT wirz-sli.o -MD -MP -MF ".deps/wirz-sli.Tpo" -c -o
wirz-sli.o wirz-sli.c; \
        then mv -f ".deps/wirz-sli.Tpo" ".deps/wirz-sli.Po"; else rm -f
".deps/wirz-sli.Tpo"; exit 1; fi
serialVFD_io.c: In function `serialVFD_write_parallel':
port.h:300: error: impossible constraint in `asm'
port.h:293: error: impossible constraint in `asm'
port.h:300: error: impossible constraint in `asm'
port.h:293: error: impossible constraint in `asm'
port.h:300: error: impossible constraint in `asm'
port.h:293: error: impossible constraint in `asm'
port.h:292: warning: 'value' might be used uninitialized in this function
make[3]: *** [serialVFD_io.o] Error 1

(In reply to comment #12)
> Is driver exclusion per-arch possible?

It is possible, but the asm error should be fixed in lcdproc-0.5.1-r1 now.
Please reply at bug #151308 whether it works.

As for the stable ppc not working, this is caused by bug #146407.

(In reply to comment #11)
> Marked 0.5.0-r2 ~ppc, but it only compiles in my ~ppc chroot, ppc stable fails
> with the error above.  0.5.1 fails in both chroots.

CC'ing ppc again as we still could use ~ppc for 0.5.1-r1. Thanks!

Keyworded ~sparc, tested with a mx212 and works fine.

sparc, can you use.mask the following stuff on the selinux/sparc profile as
well?

app-misc/lcdproc-0.5.1-r1: depends ~sparc: unsolvable
selinux/sparc/sparc64/2006.1, solutions: [ >=dev-libs/libg15render-1.1.1,
app-misc/g15daemon, dev-libs/libg15, app-misc/glcdprocdriver,
dev-embedded/libftdi, app-misc/graphlcd-base ]
app-misc/lcdproc-0.5.1-r1: rdepends ~sparc: unsolvable
selinux/sparc/sparc64/2006.1, solutions: [ >=dev-libs/libg15render-1.1.1,
app-misc/g15daemon, dev-libs/libg15, app-misc/glcdprocdriver,
dev-embedded/libftdi, app-misc/graphlcd-base ]

I'd call someone from hardened/selinux, we don't really touch those profiles.

sparc, could you duplicate your use.mask's for the USE_EXPAND'ed versions. But
please do not delete the current masks, as they are still needed for old ebuild
versions that do not get updated:
* ula200 -> lcd_devices_ula200
* graphlcd -> lcd_devices_graphlcd
* g15 -> lcd_devices_g15.

Also, as Jakub pointed out, please copy the six masks also to selinux/sparc

Compiles on ppc again, so marking ~ppc and reassigning to lcd@g.o

assinging to sparc, for the profile changes.

SPARC profiles done, but AGAIN as i stated before, we don't touch
hardened/selinux profiles.

Created an attachment (id=109891) [details]
qemu-0.9.0-nopiessp.patch

This could be done via a sed statement also. It just forces ssp and pie to 
be disabled everywhere. The op.c was not inheriting the cflags fully 
when it was built.

(From update of attachment 109891 [details])
ooops wrong bug ignore last comment please.

Jokey got the OK to fix this from pebenito, so this bug is finally closed.

I removed the use.mask's from the sparc and selinux/sparc profiles for
   ula200, graphlcd and g15
since they do not exist as normal use flags anymore, and the lcd_devices_* ones
are still masked. The last version using these flags went away some time ago.