Message ID | 1339686873-5528-1-git-send-email-afaerber@suse.de |
---|---|
State | New |
Headers | show |
Am 14.06.2012 17:14, schrieb Andreas Färber: > Commit ec5b06d (configure: ensure directory exists when creating symlinks) > moved the creation of directories into the symlink() function but forgot > the case where no symlink is created. > > This leads to build errors on arm Linux due to -I../linux-headers. > > Unbreak the build on arm Linux by reverting part of that commit. Er, "Unbreak the build by..." would suffice, no need to double-duplicate. ;) /-F > Signed-off-by: Andreas Färber <afaerber@suse.de> > --- > configure | 1 + > 1 file changed, 1 insertion(+) > > diff --git a/configure b/configure > index c2366ee..ba8cc7c 100755 > --- a/configure > +++ b/configure > @@ -3462,6 +3462,7 @@ done > > # use included Linux headers > if test "$linux" = "yes" ; then > + mkdir -p linux-headers > case "$cpu" in > i386|x86_64) > symlink "$source_path/linux-headers/asm-x86" linux-headers/asm
On 14 June 2012 16:14, Andreas Färber <afaerber@suse.de> wrote: > Commit ec5b06d (configure: ensure directory exists when creating symlinks) > moved the creation of directories into the symlink() function but forgot > the case where no symlink is created. > > This leads to build errors on arm Linux due to -I../linux-headers. > > Unbreak the build on arm Linux by reverting part of that commit. I think it would be better to just revert ec5b06d completely. The comment documenting the symlink() function says "# symbolically link $1 to $2. Portable version of "ln -sf" and "ln -sf" doesn't create the target directory, so symlink() should not either. -- PMM
Am 19.06.2012 17:18, schrieb Peter Maydell: > On 14 June 2012 16:14, Andreas Färber <afaerber@suse.de> wrote: >> Commit ec5b06d (configure: ensure directory exists when creating symlinks) >> moved the creation of directories into the symlink() function but forgot >> the case where no symlink is created. >> >> This leads to build errors on arm Linux due to -I../linux-headers. >> >> Unbreak the build on arm Linux by reverting part of that commit. > > I think it would be better to just revert ec5b06d completely. > The comment documenting the symlink() function says > "# symbolically link $1 to $2. Portable version of "ln -sf" > and "ln -sf" doesn't create the target directory, so symlink() > should not either. I don't mind either way as long as the build gets fixed. Anthony? /-F
On 19 June 2012 16:24, Andreas Färber <afaerber@suse.de> wrote: > Am 19.06.2012 17:18, schrieb Peter Maydell: >> On 14 June 2012 16:14, Andreas Färber <afaerber@suse.de> wrote: >>> Commit ec5b06d (configure: ensure directory exists when creating symlinks) >>> moved the creation of directories into the symlink() function but forgot >>> the case where no symlink is created. >>> >>> This leads to build errors on arm Linux due to -I../linux-headers. >>> >>> Unbreak the build on arm Linux by reverting part of that commit. >> >> I think it would be better to just revert ec5b06d completely. >> The comment documenting the symlink() function says >> "# symbolically link $1 to $2. Portable version of "ln -sf" >> and "ln -sf" doesn't create the target directory, so symlink() >> should not either. > > I don't mind either way as long as the build gets fixed. Anthony? It turns out that there are later changes to configure which rely on the auto-creation of the directory, so simple reversion isn't sufficient. Maybe your patch is the simplest fix after all. -- PMM
On 14 June 2012 16:14, Andreas Färber <afaerber@suse.de> wrote: > Commit ec5b06d (configure: ensure directory exists when creating symlinks) > moved the creation of directories into the symlink() function but forgot > the case where no symlink is created. > > This leads to build errors on arm Linux due to -I../linux-headers. > > Unbreak the build on arm Linux by reverting part of that commit. > > Signed-off-by: Andreas Färber <afaerber@suse.de> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> I guess we get to have the usual argument about how build fixes get committed now :-) -- PMM
On 06/14/2012 10:14 AM, Andreas Färber wrote: > Commit ec5b06d (configure: ensure directory exists when creating symlinks) > moved the creation of directories into the symlink() function but forgot > the case where no symlink is created. > > This leads to build errors on arm Linux due to -I../linux-headers. > > Unbreak the build on arm Linux by reverting part of that commit. > > Signed-off-by: Andreas Färber<afaerber@suse.de> Applied. Thanks. Regards, Anthony Liguori > --- > configure | 1 + > 1 file changed, 1 insertion(+) > > diff --git a/configure b/configure > index c2366ee..ba8cc7c 100755 > --- a/configure > +++ b/configure > @@ -3462,6 +3462,7 @@ done > > # use included Linux headers > if test "$linux" = "yes" ; then > + mkdir -p linux-headers > case "$cpu" in > i386|x86_64) > symlink "$source_path/linux-headers/asm-x86" linux-headers/asm
diff --git a/configure b/configure index c2366ee..ba8cc7c 100755 --- a/configure +++ b/configure @@ -3462,6 +3462,7 @@ done # use included Linux headers if test "$linux" = "yes" ; then + mkdir -p linux-headers case "$cpu" in i386|x86_64) symlink "$source_path/linux-headers/asm-x86" linux-headers/asm
Commit ec5b06d (configure: ensure directory exists when creating symlinks) moved the creation of directories into the symlink() function but forgot the case where no symlink is created. This leads to build errors on arm Linux due to -I../linux-headers. Unbreak the build on arm Linux by reverting part of that commit. Signed-off-by: Andreas Färber <afaerber@suse.de> --- configure | 1 + 1 file changed, 1 insertion(+)