Message ID | 20200802203421.5959-1-kuba@kadziolka.net |
---|---|
State | Accepted |
Headers | show |
Series | [bug#42681,core-updates] gnu: bootstrap-gcc: Don't use CPATH search paths. | expand |
Context | Check | Description |
---|---|---|
cbaines/comparison | success | View comparision |
cbaines/git branch | success | View Git branch |
cbaines/applying patch | success | View Laminar job |
On Sun, Aug 02, 2020 at 10:34:21PM +0200, Jakub Kądziołka wrote: > This updates the bootstrap-gcc to use the same search path strategy as > the main compilers, making the bootstrap environment more similar across > architectures, and fixing the bootstrap path on non-x86 architectures > again. > > This is a follow up to commit 7cde865f12dd67e55946a52a21103e71f447eced. > > * gnu/packages/bootstrap.scm (%bootstrap-gcc)[native-search-paths]: Use > C_INCLUDE_PATH and CPLUS_INCLUDE_PATH instead of CPATH. > --- > > Note: because I don't have any non-x86 machines on hand, I haven't > actually tested the patch yet. > > Efraim, you have helped me investigate this with your ARM board, perhaps > you could test this patch? > > Regards, > Jakub Kądziołka > gnu/packages/bootstrap.scm | 5 ++++- > 1 file changed, 4 insertions(+), 1 deletion(-) > > diff --git a/gnu/packages/bootstrap.scm b/gnu/packages/bootstrap.scm > index c39e60ec8b..fd6c2e099c 100644 > --- a/gnu/packages/bootstrap.scm > +++ b/gnu/packages/bootstrap.scm > @@ -740,8 +740,11 @@ exec ~a/bin/.gcc-wrapped -B~a/lib \ > "1m5miqkyng45l745n0sfafdpjkqv9225xf44jqkygwsipj2cv9ks"))))))))) > (native-search-paths > (list (search-path-specification > - (variable "CPATH") > + (variable "C_INCLUDE_PATH") > (files '("include"))) > + (search-path-specification > + (variable "CPLUS_INCLUDE_PATH") > + (files '("include/c++" "include"))) > (search-path-specification > (variable "LIBRARY_PATH") > (files '("lib" "lib64"))))) > -- > 2.28.0 > I'm building out to hello on my aarch64 board now, starting from make-boot0 apparently. Guile is only listed once so hopefully it won't take too long. I assume this doesn't break anything on x86_64?
On Mon, Aug 03, 2020 at 09:40:26AM +0300, Efraim Flashner wrote: > On Sun, Aug 02, 2020 at 10:34:21PM +0200, Jakub Kądziołka wrote: > > This updates the bootstrap-gcc to use the same search path strategy as > > the main compilers, making the bootstrap environment more similar across > > architectures, and fixing the bootstrap path on non-x86 architectures > > again. > > > > This is a follow up to commit 7cde865f12dd67e55946a52a21103e71f447eced. > > > > * gnu/packages/bootstrap.scm (%bootstrap-gcc)[native-search-paths]: Use > > C_INCLUDE_PATH and CPLUS_INCLUDE_PATH instead of CPATH. > > --- > > > > Note: because I don't have any non-x86 machines on hand, I haven't > > actually tested the patch yet. > > > > Efraim, you have helped me investigate this with your ARM board, perhaps > > you could test this patch? > > > > Regards, > > Jakub Kądziołka > > gnu/packages/bootstrap.scm | 5 ++++- > > 1 file changed, 4 insertions(+), 1 deletion(-) > > > > diff --git a/gnu/packages/bootstrap.scm b/gnu/packages/bootstrap.scm > > index c39e60ec8b..fd6c2e099c 100644 > > --- a/gnu/packages/bootstrap.scm > > +++ b/gnu/packages/bootstrap.scm > > @@ -740,8 +740,11 @@ exec ~a/bin/.gcc-wrapped -B~a/lib \ > > "1m5miqkyng45l745n0sfafdpjkqv9225xf44jqkygwsipj2cv9ks"))))))))) > > (native-search-paths > > (list (search-path-specification > > - (variable "CPATH") > > + (variable "C_INCLUDE_PATH") > > (files '("include"))) > > + (search-path-specification > > + (variable "CPLUS_INCLUDE_PATH") > > + (files '("include/c++" "include"))) > > (search-path-specification > > (variable "LIBRARY_PATH") > > (files '("lib" "lib64"))))) > > -- > > 2.28.0 > > > > I'm building out to hello on my aarch64 board now, starting from > make-boot0 apparently. Guile is only listed once so hopefully it won't > take too long. Thanks! > I assume this doesn't break anything on x86_64? I don't have hello built on current core-updates (it also wants to start from make-boot0), but I have confirmed that guix build --no-grafts -d hello returns the same derivation path before and after the commit. %bootstrap-gcc is not present at all in the package graph on x86. Regards, Jakub Kądziołka
On Mon, Aug 03, 2020 at 01:03:52PM +0200, Jakub Kądziołka wrote: > On Mon, Aug 03, 2020 at 09:40:26AM +0300, Efraim Flashner wrote: > > On Sun, Aug 02, 2020 at 10:34:21PM +0200, Jakub Kądziołka wrote: > > > This updates the bootstrap-gcc to use the same search path strategy as > > > the main compilers, making the bootstrap environment more similar across > > > architectures, and fixing the bootstrap path on non-x86 architectures > > > again. > > > > > > This is a follow up to commit 7cde865f12dd67e55946a52a21103e71f447eced. > > > > > > * gnu/packages/bootstrap.scm (%bootstrap-gcc)[native-search-paths]: Use > > > C_INCLUDE_PATH and CPLUS_INCLUDE_PATH instead of CPATH. > > > --- > > > > > > Note: because I don't have any non-x86 machines on hand, I haven't > > > actually tested the patch yet. > > > > > > Efraim, you have helped me investigate this with your ARM board, perhaps > > > you could test this patch? > > > > > > Regards, > > > Jakub Kądziołka > > > gnu/packages/bootstrap.scm | 5 ++++- > > > 1 file changed, 4 insertions(+), 1 deletion(-) > > > > > > diff --git a/gnu/packages/bootstrap.scm b/gnu/packages/bootstrap.scm > > > index c39e60ec8b..fd6c2e099c 100644 > > > --- a/gnu/packages/bootstrap.scm > > > +++ b/gnu/packages/bootstrap.scm > > > @@ -740,8 +740,11 @@ exec ~a/bin/.gcc-wrapped -B~a/lib \ > > > "1m5miqkyng45l745n0sfafdpjkqv9225xf44jqkygwsipj2cv9ks"))))))))) > > > (native-search-paths > > > (list (search-path-specification > > > - (variable "CPATH") > > > + (variable "C_INCLUDE_PATH") > > > (files '("include"))) > > > + (search-path-specification > > > + (variable "CPLUS_INCLUDE_PATH") > > > + (files '("include/c++" "include"))) > > > (search-path-specification > > > (variable "LIBRARY_PATH") > > > (files '("lib" "lib64"))))) > > > -- > > > 2.28.0 > > > > > > > I'm building out to hello on my aarch64 board now, starting from > > make-boot0 apparently. Guile is only listed once so hopefully it won't > > take too long. > > Thanks! > > > I assume this doesn't break anything on x86_64? > > I don't have hello built on current core-updates (it also wants to start > from make-boot0), but I have confirmed that guix build --no-grafts -d > hello returns the same derivation path before and after the commit. > %bootstrap-gcc is not present at all in the package graph on x86. > > Regards, > Jakub Kądziołka I was able to successfully build all the way out to hello on aarch64-linux.
On Tue, Aug 04, 2020 at 09:00:06AM +0300, Efraim Flashner wrote: > On Mon, Aug 03, 2020 at 01:03:52PM +0200, Jakub Kądziołka wrote: > > On Mon, Aug 03, 2020 at 09:40:26AM +0300, Efraim Flashner wrote: > > > On Sun, Aug 02, 2020 at 10:34:21PM +0200, Jakub Kądziołka wrote: > > > I'm building out to hello on my aarch64 board now, starting from > > > make-boot0 apparently. Guile is only listed once so hopefully it won't > > > take too long. > > > > Thanks! > > > > > I assume this doesn't break anything on x86_64? > > > > I don't have hello built on current core-updates (it also wants to start > > from make-boot0), but I have confirmed that guix build --no-grafts -d > > hello returns the same derivation path before and after the commit. > > %bootstrap-gcc is not present at all in the package graph on x86. > > > > Regards, > > Jakub Kądziołka > > I was able to successfully build all the way out to hello on > aarch64-linux. This is great news! Patch pushed. Regards, Jakub Kądziołka
On Tue, Aug 04, 2020 at 02:34:35PM +0200, Jakub Kądziołka wrote: > On Tue, Aug 04, 2020 at 09:00:06AM +0300, Efraim Flashner wrote: > > On Mon, Aug 03, 2020 at 01:03:52PM +0200, Jakub Kądziołka wrote: > > > On Mon, Aug 03, 2020 at 09:40:26AM +0300, Efraim Flashner wrote: > > > > On Sun, Aug 02, 2020 at 10:34:21PM +0200, Jakub Kądziołka wrote: > > > > I'm building out to hello on my aarch64 board now, starting from > > > > make-boot0 apparently. Guile is only listed once so hopefully it won't > > > > take too long. > > > > > > Thanks! > > > > > > > I assume this doesn't break anything on x86_64? > > > > > > I don't have hello built on current core-updates (it also wants to start > > > from make-boot0), but I have confirmed that guix build --no-grafts -d > > > hello returns the same derivation path before and after the commit. > > > %bootstrap-gcc is not present at all in the package graph on x86. > > > > > > > I was able to successfully build all the way out to hello on > > aarch64-linux. > > This is great news! Patch pushed. > Continued testing shows that this also worked on my aarch64 board when building for armhf.
diff --git a/gnu/packages/bootstrap.scm b/gnu/packages/bootstrap.scm index c39e60ec8b..fd6c2e099c 100644 --- a/gnu/packages/bootstrap.scm +++ b/gnu/packages/bootstrap.scm @@ -740,8 +740,11 @@ exec ~a/bin/.gcc-wrapped -B~a/lib \ "1m5miqkyng45l745n0sfafdpjkqv9225xf44jqkygwsipj2cv9ks"))))))))) (native-search-paths (list (search-path-specification - (variable "CPATH") + (variable "C_INCLUDE_PATH") (files '("include"))) + (search-path-specification + (variable "CPLUS_INCLUDE_PATH") + (files '("include/c++" "include"))) (search-path-specification (variable "LIBRARY_PATH") (files '("lib" "lib64")))))