Skip to contentSkip to conversation

package/libxml-parser-perl: make host build use correct compiler

PATCH Superseded Norbert Lange 2 participants 4 messages Jun 5, 2020 – Jun 7, 2020

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 package/libxml-parser-perl: make host build use correct compiler

This package uses gcc filename without path, which breaks the host build if host and target compiler have the same filename.
We pack the HOST variables as override so absolute paths are used for the host tools.

---
 package/libxml-parser-perl/libxml-parser-perl.mk | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
package/libxml-parser-perl/libxml-parser-perl.mk +1 -1
index fcde5fc939..100c49633c 100644
@@ -25,7 +25,7 @@ define HOST_LIBXML_PARSER_PERL_CONFIGURE_CMDS
 endef
 
 define HOST_LIBXML_PARSER_PERL_BUILD_CMDS
-	$(HOST_MAKE_ENV) $(MAKE) -C $(@D)
+	$(HOST_MAKE_ENV) $(MAKE) $(HOST_CONFIGURE_OPTS) -C $(@D)
 endef
 
 define HOST_LIBXML_PARSER_PERL_INSTALL_CMDS
Signature · 1 line
2.26.2
#1 lore raw

PATCH 1/1 package/libxml-parser-perl: make host build use correct compiler

Quoted 2 lines · On Sat, 6 Jun 2020 00:58:12 +0200 Norbert Lange <[email protected]> wrote:

This package uses gcc filename without path, which breaks the host build if host and target compiler have the same filename.

How is it possible for the host and target compilers to have the same file name ?

Quoted 2 lines

We pack the HOST variables as override so absolute paths are used for the host tools.

I'm not sure what you mean by "we pack the HOST variables".

Note: I do understand that the patch will ensure that the compiler point by $(HOSTCC) will be used, instead of whatever default the libxml-parser-perl build system thinks it should use. But your commit log is very confusing.

Also, your Signed-off-by is missing.

Could you resend, with a clarified explanation, and your SoB added ?

Thanks!

Thomas

Signature · 3 lines
Thomas Petazzoni, CTO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
#2 lore raw

PATCH 1/1 package/libxml-parser-perl: make host build use correct compiler

Am Sa., 6. Juni 2020 um 22:04 Uhr schrieb Thomas Petazzoni <[email protected]>:

Quoted 9 lines
Quoted 2 lines · On Sat, 6 Jun 2020 00:58:12 +0200 Norbert Lange <[email protected]> wrote:

This package uses gcc filename without path, which breaks the host build if host and target compiler have the same filename.

How is it possible for the host and target compilers to have the same file name ?

I use crosstool for a x86_64 target compiler, same architecture as my desktop pc.

Quoted 4 lines
Quoted 3 lines

We pack the HOST variables as override so absolute paths are used for the host tools.

Quoted 8 lines

I'm not sure what you mean by "we pack the HOST variables".

Note: I do understand that the patch will ensure that the compiler point by $(HOSTCC) will be used, instead of whatever default the libxml-parser-perl build system thinks it should use. But your commit log is very confusing.

Also, your Signed-off-by is missing.

Was using git send-email blindly without checking, sorry.

Quoted 2 lines

Could you resend, with a clarified explanation, and your SoB added ?

Done.

On an not entirely unrelated note, I got troubles getting into patchwork, did reset my PW 2-3 time, still can't log in. Is there some hidden 8 letter limit or similar?

Norbert

#3 lore raw

PATCH 1/1 package/libxml-parser-perl: make host build use correct compiler

Quoted 5 lines · On Sun, 7 Jun 2020 00:29:59 +0200 Norbert Lange <[email protected]> wrote:
Quoted 2 lines

How is it possible for the host and target compilers to have the same file name ?

I use crosstool for a x86_64 target compiler, same architecture as my desktop pc.

That is still weird: the native compiler on your system is normally called "gcc", while the compiler in your toolchain is generally named x86_64-linux-gcc or some variation on that.

Even though your patch is definitely correct, I'd like to understand how you fall into this situation. Indeed, I'm pretty sure several other host packages might be affected, and if this is a valid situation, we should see if there's a way to test it in our autobuilders.

Quoted 3 lines

On an not entirely unrelated note, I got troubles getting into patchwork, did reset my PW 2-3 time, still can't log in. Is there some hidden 8 letter limit or similar?

I don't think there is such a limitation. This patchwork instance is not always super stable, yesterday we had 500 errors for a few moments for example. If you have issues, we can contact the admins of the instance and see if they can have a look.

Best regards,

Thomas

Signature · 3 lines
Thomas Petazzoni, CTO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com