Patchwork [v3] toolchain-external: Fix EABIhf check

login
register
mail settings
Submitter Thomas De Schampheleire
Date May 6, 2014, 11:25 a.m.
Message ID <2531b66f78bfdb046ddb.1399375540@argentina>
Download mbox | patch
Permalink /patch/346134/
State Superseded
Headers show

Comments

Thomas De Schampheleire - May 6, 2014, 11:25 a.m.
From: Stefan Sørensen <stefan.sorensen@spectralink.com>

Currently the check for EABI/EABIhf toolchains looks for the
Tag_ABI_VFP_args attribute in the crt1.o file which gcc adds in an
EABIhf toolchain.
In uClibc, however, crt1.o is not compiled from c but assembly, so the
Tag_ABI_VFP_args attribute is not added in the object file. This causes
the EABIhf check in the external toolchain logic to fail for
uClibc-based toolchains.

Fix by compiling a dummy .c file and trying to link the object against the
C library. Since it is impossible to mix EABI and EABIhf code, a mismatch
between the buildroot and toolchain ABI settings will be detected during
this link step.

Fixes bug #6842: https://bugs.busybox.net/show_bug.cgi?id=6842

Signed-off-by: Stefan Sørensen <stefan.sorensen@spectralink.com>
[ThomasDS: do full link iso readelf test, update commit message]
Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>

---
v3: (ThomasDS)
  - v2 properly recognized a uClibc EABIhf toolchain, but failed to detect a
    mismatch with buildroot settings (detected by ThomasP). Fix by
    performing an actual link step to detect any incompatibility.
v2: (ThomasDS)
  - avoid creating a temporary source file
  - update commit message

 toolchain/helpers.mk |  18 ++++++------------
 1 files changed, 6 insertions(+), 12 deletions(-)
Arnout Vandecappelle - May 8, 2014, 10:35 p.m.
On 06/05/14 13:25, Thomas De Schampheleire wrote:
> From: Stefan Sørensen <stefan.sorensen@spectralink.com>
> 
> Currently the check for EABI/EABIhf toolchains looks for the
> Tag_ABI_VFP_args attribute in the crt1.o file which gcc adds in an
> EABIhf toolchain.
> In uClibc, however, crt1.o is not compiled from c but assembly, so the
> Tag_ABI_VFP_args attribute is not added in the object file. This causes
> the EABIhf check in the external toolchain logic to fail for
> uClibc-based toolchains.
> 
> Fix by compiling a dummy .c file and trying to link the object against the
> C library. Since it is impossible to mix EABI and EABIhf code, a mismatch
> between the buildroot and toolchain ABI settings will be detected during
> this link step.
> 
> Fixes bug #6842: https://bugs.busybox.net/show_bug.cgi?id=6842
> 
> Signed-off-by: Stefan Sørensen <stefan.sorensen@spectralink.com>
> [ThomasDS: do full link iso readelf test, update commit message]
> Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
> 
> ---
> v3: (ThomasDS)
>   - v2 properly recognized a uClibc EABIhf toolchain, but failed to detect a
>     mismatch with buildroot settings (detected by ThomasP). Fix by
>     performing an actual link step to detect any incompatibility.
> v2: (ThomasDS)
>   - avoid creating a temporary source file
>   - update commit message
> 
>  toolchain/helpers.mk |  18 ++++++------------
>  1 files changed, 6 insertions(+), 12 deletions(-)
> 
> diff --git a/toolchain/helpers.mk b/toolchain/helpers.mk
> --- a/toolchain/helpers.mk
> +++ b/toolchain/helpers.mk
> @@ -297,20 +297,14 @@ check_arm_abi = \
>  		echo "External toolchain uses the unsuported OABI" ; \
>  		exit 1 ; \
>  	fi ; \
> -	EXT_TOOLCHAIN_CRT1=`LANG=C $${__CROSS_CC} -print-file-name=crt1.o` ; \
> -	if $${__CROSS_READELF} -A $${EXT_TOOLCHAIN_CRT1} | grep -q "Tag_ABI_VFP_args:" ; then \
> -		EXT_TOOLCHAIN_ABI="eabihf" ; \
> -	else \
> -		EXT_TOOLCHAIN_ABI="eabi" ; \
> -	fi ; \
> -	if [ "$(BR2_ARM_EABI)" = "y" -a "$${EXT_TOOLCHAIN_ABI}" = "eabihf" ] ; then \
> -		echo "Incorrect ABI setting: EABI selected, but toolchain uses EABIhf" ; \
> +	TEMP_OBJ_FILE=`mktemp --tmpdir --suffix=.o`; \
> +	if ! echo 'int main(void) {}' | $${__CROSS_CC} -x c -o $${TEMP_OBJ_FILE} /dev/null - 2>/dev/null; then \

 You can write directly to /dev/null instead of going through TEMP_OBJ_FILE, no?

> +		abistr_$(BR2_ARM_EABI)='EABI'; \
> +		abistr_$(BR2_ARM_EABIHF)='EABIhf'; \
> +		echo "Incorrect ABI setting: $${abistr_y} selected, but toolchain is incompatible"; \

 Nice approach! But to keep the original output, you could use:

$${abistr_y} selected, but toolchain uses $${abistr_}


 Regards,
 Arnout

>  		exit 1 ; \
>  	fi ; \
> -	if [ "$(BR2_ARM_EABIHF)" = "y" -a "$${EXT_TOOLCHAIN_ABI}" = "eabi" ] ; then \
> -		echo "Incorrect ABI setting: EABIhf selected, but toolchain uses EABI" ; \
> -		exit 1 ; \
> -	fi
> +	rm -f $${TEMP_OBJ_FILE}
>  
>  #
>  # Check that the external toolchain supports C++
> _______________________________________________
> buildroot mailing list
> buildroot@busybox.net
> http://lists.busybox.net/mailman/listinfo/buildroot
>
Thomas De Schampheleire - May 9, 2014, 7:23 a.m.
Hi Arnout,

On Fri, May 9, 2014 at 12:35 AM, Arnout Vandecappelle <arnout@mind.be> wrote:
> On 06/05/14 13:25, Thomas De Schampheleire wrote:
>> From: Stefan Sørensen <stefan.sorensen@spectralink.com>
>>
>> Currently the check for EABI/EABIhf toolchains looks for the
>> Tag_ABI_VFP_args attribute in the crt1.o file which gcc adds in an
>> EABIhf toolchain.
>> In uClibc, however, crt1.o is not compiled from c but assembly, so the
>> Tag_ABI_VFP_args attribute is not added in the object file. This causes
>> the EABIhf check in the external toolchain logic to fail for
>> uClibc-based toolchains.
>>
>> Fix by compiling a dummy .c file and trying to link the object against the
>> C library. Since it is impossible to mix EABI and EABIhf code, a mismatch
>> between the buildroot and toolchain ABI settings will be detected during
>> this link step.
>>
>> Fixes bug #6842: https://bugs.busybox.net/show_bug.cgi?id=6842
>>
>> Signed-off-by: Stefan Sørensen <stefan.sorensen@spectralink.com>
>> [ThomasDS: do full link iso readelf test, update commit message]
>> Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
>>
>> ---
>> v3: (ThomasDS)
>>   - v2 properly recognized a uClibc EABIhf toolchain, but failed to detect a
>>     mismatch with buildroot settings (detected by ThomasP). Fix by
>>     performing an actual link step to detect any incompatibility.
>> v2: (ThomasDS)
>>   - avoid creating a temporary source file
>>   - update commit message
>>
>>  toolchain/helpers.mk |  18 ++++++------------
>>  1 files changed, 6 insertions(+), 12 deletions(-)
>>
>> diff --git a/toolchain/helpers.mk b/toolchain/helpers.mk
>> --- a/toolchain/helpers.mk
>> +++ b/toolchain/helpers.mk
>> @@ -297,20 +297,14 @@ check_arm_abi = \
>>               echo "External toolchain uses the unsuported OABI" ; \
>>               exit 1 ; \
>>       fi ; \
>> -     EXT_TOOLCHAIN_CRT1=`LANG=C $${__CROSS_CC} -print-file-name=crt1.o` ; \
>> -     if $${__CROSS_READELF} -A $${EXT_TOOLCHAIN_CRT1} | grep -q "Tag_ABI_VFP_args:" ; then \
>> -             EXT_TOOLCHAIN_ABI="eabihf" ; \
>> -     else \
>> -             EXT_TOOLCHAIN_ABI="eabi" ; \
>> -     fi ; \
>> -     if [ "$(BR2_ARM_EABI)" = "y" -a "$${EXT_TOOLCHAIN_ABI}" = "eabihf" ] ; then \
>> -             echo "Incorrect ABI setting: EABI selected, but toolchain uses EABIhf" ; \
>> +     TEMP_OBJ_FILE=`mktemp --tmpdir --suffix=.o`; \
>> +     if ! echo 'int main(void) {}' | $${__CROSS_CC} -x c -o $${TEMP_OBJ_FILE} /dev/null - 2>/dev/null; then \
>
>  You can write directly to /dev/null instead of going through TEMP_OBJ_FILE, no?

Ah yes. When we were still using the readelf approach I tried avoiding
the temporary file by using 'gcc -o -' so that I could pipe it into
readelf, but gcc did not accept that.
When changing approach I did not realize I could remove it, thanks.

I also noticed now that there is an extra /dev/null on this line as
input file, which is not needed.

>
>> +             abistr_$(BR2_ARM_EABI)='EABI'; \
>> +             abistr_$(BR2_ARM_EABIHF)='EABIhf'; \
>> +             echo "Incorrect ABI setting: $${abistr_y} selected, but toolchain is incompatible"; \
>
>  Nice approach! But to keep the original output, you could use:
>
> $${abistr_y} selected, but toolchain uses $${abistr_}
>

This did cross my mind, but it assumes that there are only two
possibilities: EABI and EABIhf. Suppose that a new EABI pops up. Or
suppose the compilation/link step fails for a completely different
reason. Then we have told the user that their toolchain is EABI or
EABIhf while it is not true.
So here I choice correctness over completeness, but if you prefer I
can change it. Do let me know if you still think we should add
$${abistr_} to the message.

Thanks,
Thomas
Arnout Vandecappelle - May 9, 2014, 7:53 p.m.
On 09/05/14 09:23, Thomas De Schampheleire wrote:
> Hi Arnout,
> 
> On Fri, May 9, 2014 at 12:35 AM, Arnout Vandecappelle <arnout@mind.be> wrote:
>> On 06/05/14 13:25, Thomas De Schampheleire wrote:
[snip]
>>
>>> +             abistr_$(BR2_ARM_EABI)='EABI'; \
>>> +             abistr_$(BR2_ARM_EABIHF)='EABIhf'; \
>>> +             echo "Incorrect ABI setting: $${abistr_y} selected, but toolchain is incompatible"; \
>>
>>  Nice approach! But to keep the original output, you could use:
>>
>> $${abistr_y} selected, but toolchain uses $${abistr_}
>>
> 
> This did cross my mind, but it assumes that there are only two
> possibilities: EABI and EABIhf. Suppose that a new EABI pops up. Or
> suppose the compilation/link step fails for a completely different
> reason. Then we have told the user that their toolchain is EABI or
> EABIhf while it is not true.
> So here I choice correctness over completeness, but if you prefer I
> can change it. Do let me know if you still think we should add
> $${abistr_} to the message.

 Good point, let's say "incompatible" then.

 Regards,
 Arnout

Patch

diff --git a/toolchain/helpers.mk b/toolchain/helpers.mk
--- a/toolchain/helpers.mk
+++ b/toolchain/helpers.mk
@@ -297,20 +297,14 @@  check_arm_abi = \
 		echo "External toolchain uses the unsuported OABI" ; \
 		exit 1 ; \
 	fi ; \
-	EXT_TOOLCHAIN_CRT1=`LANG=C $${__CROSS_CC} -print-file-name=crt1.o` ; \
-	if $${__CROSS_READELF} -A $${EXT_TOOLCHAIN_CRT1} | grep -q "Tag_ABI_VFP_args:" ; then \
-		EXT_TOOLCHAIN_ABI="eabihf" ; \
-	else \
-		EXT_TOOLCHAIN_ABI="eabi" ; \
-	fi ; \
-	if [ "$(BR2_ARM_EABI)" = "y" -a "$${EXT_TOOLCHAIN_ABI}" = "eabihf" ] ; then \
-		echo "Incorrect ABI setting: EABI selected, but toolchain uses EABIhf" ; \
+	TEMP_OBJ_FILE=`mktemp --tmpdir --suffix=.o`; \
+	if ! echo 'int main(void) {}' | $${__CROSS_CC} -x c -o $${TEMP_OBJ_FILE} /dev/null - 2>/dev/null; then \
+		abistr_$(BR2_ARM_EABI)='EABI'; \
+		abistr_$(BR2_ARM_EABIHF)='EABIhf'; \
+		echo "Incorrect ABI setting: $${abistr_y} selected, but toolchain is incompatible"; \
 		exit 1 ; \
 	fi ; \
-	if [ "$(BR2_ARM_EABIHF)" = "y" -a "$${EXT_TOOLCHAIN_ABI}" = "eabi" ] ; then \
-		echo "Incorrect ABI setting: EABIhf selected, but toolchain uses EABI" ; \
-		exit 1 ; \
-	fi
+	rm -f $${TEMP_OBJ_FILE}
 
 #
 # Check that the external toolchain supports C++