seabios build failure in xen tree

seabios build failure in xen tree

[Christoph Egger]

Adding seabios ML.

On 01/26/12 16:06, Ian Campbell wrote:
 > On Thu, 2012-01-26 at 15:01 +0000, Christoph Egger wrote:
 >> On 01/26/12 15:51, Ian Campbell wrote:
 >>> On Thu, 2012-01-26 at 14:35 +0000, Christoph Egger wrote:
 >>>> gmake[6]: Entering directory tools/firmware/seabios-dir-remote
 >>>>      Building ld scripts (version "1.6.3.1-20120126_152501")
 >>>> env: python: No such file or directory
 >>>> gmake[6]: *** [out/romlayout16.lds] Error 127
 >>>>
 >>>>
 >>>> The python scripts must be invoked with $(PYTHON) as done
 >>>> throughout the build system.
 >>>
 >>> SeaBIOS uses:
 >>> $ rgrep -i python tools/firmware/seabios-dir
 >>> tools/firmware/seabios-dir/tools/transdump.py:#!/usr/bin/env python
 >>> tools/firmware/seabios-dir/tools/buildrom.py:#!/usr/bin/env python
 >>> tools/firmware/seabios-dir/tools/checkstack.py:#!/usr/bin/env python
 >>> tools/firmware/seabios-dir/tools/encodeint.py:#!/usr/bin/env python
 >>> tools/firmware/seabios-dir/tools/layoutrom.py:#!/usr/bin/env python
 >>> tools/firmware/seabios-dir/tools/checkrom.py:#!/usr/bin/env python
 >>> tools/firmware/seabios-dir/tools/readserial.py:#!/usr/bin/env python
 >>> tools/firmware/seabios-dir/tools/readserial.py:Or: apt-get install 
python-serial
 >>> tools/firmware/seabios-dir/tools/checksum.py:#!/usr/bin/env python
 >>>
 >>> Does this not work? Should python be on your $PATH?
 >>
 >> It is. But the python binary on NetBSD's pkgsrc
 >> is called python<version>  to allow concurrent installations of 
different
 >> python versions.
 >> So I can have: python2.5, python2.6, python2.7, python3.1, etc.
 >
 > There is no current "python" referring to the default version?
 >
 > I think this is something you will need to work out with the SeaBIOS
 > upstream.
 >
 > Ian.
 >
 >>
 >> Christoph
 >>
 >>>
 >>> Since SeaBIOS is third party code we are not so much at liberty to make
 >>> the same sorts of policy decisions as we would for our own code.
 >>>
 >>> You could perhaps attempt to work around this by invoking the recursive
 >>> make into the seabios directory with PATH=$(dir-only $(PYTHON):$PATH or
 >>> whatever the appropriate runes are.
 >>>
 >>> Or you could try posting a patch to SeaBIOS-devel, I don't know what
 >>> their thinking on this sort of thing is.

I see.

What python version does SeaBIOS require?
I have python 2.5 installed.

I manually created a python symlink to that version and then I get this 
failure:

gmake[6]: Entering directory 'tools/firmware/seabios-dir-remote'
   Building ld scripts (version "1.6.3.1-20120127_151243")
Fixed space: 0xe05b-0x10000  total: 8101  slack: 5 Percent slack: 0.1%
16bit size:           46336
32bit segmented size: 2005
32bit flat size:      14699
32bit flat init size: 53888
Traceback (most recent call last):
   File "./tools/layoutrom.py", line 579, in <module>
     main()
   File "./tools/layoutrom.py", line 576, in main
     writeLinkerScripts(sections, entrysym, genreloc, out16, out32seg, 
out32flag)
   File "./tools/layoutrom.py", line 257, in writeLinkerScripts
     + COMMONTRAILER
TypeError: int argument required
gmake[6]: *** [out/romlayout16.lds] Error 1


Christoph


-- 
---to satisfy European Law for business letters:
Advanced Micro Devices GmbH
Einsteinring 24, 85689 Dornach b. Muenchen
Geschaeftsfuehrer: Alberto Bozzo, Andrew Bowd
Sitz: Dornach, Gemeinde Aschheim, Landkreis Muenchen
Registergericht Muenchen, HRB Nr. 43632

Re: seabios build failure in xen tree

[Kevin O'Connor]

On Tue, Feb 07, 2012 at 02:35:11PM +0100, Christoph Egger wrote:
> 
> Adding seabios ML.
> 
> On 01/26/12 16:06, Ian Campbell wrote:
> > On Thu, 2012-01-26 at 15:01 +0000, Christoph Egger wrote:
> >> On 01/26/12 15:51, Ian Campbell wrote:
> >>> On Thu, 2012-01-26 at 14:35 +0000, Christoph Egger wrote:
> >>>> gmake[6]: Entering directory tools/firmware/seabios-dir-remote
> >>>>      Building ld scripts (version "1.6.3.1-20120126_152501")
> >>>> env: python: No such file or directory
> >>>> gmake[6]: *** [out/romlayout16.lds] Error 127
> >>>>
> >>>>
> >>>> The python scripts must be invoked with $(PYTHON) as done
> >>>> throughout the build system.

We can change the seabios makefile to call these scripts with an
explicit $(PYTHON).  However, it's odd that there isn't a default
python on your system.

[...]
> What python version does SeaBIOS require?
> I have python 2.5 installed.

It should work with python 2.5 - there's nothing really special going
on in the script.

> I manually created a python symlink to that version and then I get
> this failure:
> 
> gmake[6]: Entering directory 'tools/firmware/seabios-dir-remote'
>   Building ld scripts (version "1.6.3.1-20120127_151243")
> Fixed space: 0xe05b-0x10000  total: 8101  slack: 5 Percent slack: 0.1%
> 16bit size:           46336
> 32bit segmented size: 2005
> 32bit flat size:      14699
> 32bit flat init size: 53888
> Traceback (most recent call last):
>   File "./tools/layoutrom.py", line 579, in <module>
>     main()
>   File "./tools/layoutrom.py", line 576, in main
>     writeLinkerScripts(sections, entrysym, genreloc, out16,
> out32seg, out32flag)
>   File "./tools/layoutrom.py", line 257, in writeLinkerScripts
>     + COMMONTRAILER
> TypeError: int argument required

That's quite odd.  This looks data related instead of python related.
Try running a "make clean" and then "make" in just the seabios
directory.  If you still see an issue, tar up the seabios "out/"
directory and mail it to me.

-Kevin

Re: [SeaBIOS] seabios build failure in xen tree

[Christoph Egger]

On 02/08/12 02:18, Kevin O'Connor wrote:
> On Tue, Feb 07, 2012 at 02:35:11PM +0100, Christoph Egger wrote:
>>
>> Adding seabios ML.
>>
>> On 01/26/12 16:06, Ian Campbell wrote:
>>> On Thu, 2012-01-26 at 15:01 +0000, Christoph Egger wrote:
>>>> On 01/26/12 15:51, Ian Campbell wrote:
>>>>> On Thu, 2012-01-26 at 14:35 +0000, Christoph Egger wrote:
>>>>>> gmake[6]: Entering directory tools/firmware/seabios-dir-remote
>>>>>>       Building ld scripts (version "1.6.3.1-20120126_152501")
>>>>>> env: python: No such file or directory
>>>>>> gmake[6]: *** [out/romlayout16.lds] Error 127
>>>>>>
>>>>>>
>>>>>> The python scripts must be invoked with $(PYTHON) as done
>>>>>> throughout the build system.
>
> We can change the seabios makefile to call these scripts with an
> explicit $(PYTHON).

Awesome!

> However, it's odd that there isn't a default python on your system.
>
> [...]
>> What python version does SeaBIOS require?
>> I have python 2.5 installed.
>
> It should work with python 2.5 - there's nothing really special going
> on in the script.
>
>> I manually created a python symlink to that version and then I get
>> this failure:
>>
>> gmake[6]: Entering directory 'tools/firmware/seabios-dir-remote'
>>    Building ld scripts (version "1.6.3.1-20120127_151243")
>> Fixed space: 0xe05b-0x10000  total: 8101  slack: 5 Percent slack: 0.1%
>> 16bit size:           46336
>> 32bit segmented size: 2005
>> 32bit flat size:      14699
>> 32bit flat init size: 53888
>> Traceback (most recent call last):
>>    File "./tools/layoutrom.py", line 579, in<module>
>>      main()
>>    File "./tools/layoutrom.py", line 576, in main
>>      writeLinkerScripts(sections, entrysym, genreloc, out16,
>> out32seg, out32flag)
>>    File "./tools/layoutrom.py", line 257, in writeLinkerScripts
>>      + COMMONTRAILER
>> TypeError: int argument required
>
> That's quite odd.  This looks data related instead of python related.
> Try running a "make clean" and then "make" in just the seabios
> directory.  If you still see an issue, tar up the seabios "out/"
> directory and mail it to me.

It failed the same way.
I sent you the tarball offlist due its size.

Christoph


-- 
---to satisfy European Law for business letters:
Advanced Micro Devices GmbH
Einsteinring 24, 85689 Dornach b. Muenchen
Geschaeftsfuehrer: Alberto Bozzo, Andrew Bowd
Sitz: Dornach, Gemeinde Aschheim, Landkreis Muenchen
Registergericht Muenchen, HRB Nr. 43632

Re: seabios build failure in xen tree

[Kevin O'Connor]

[-- Attachment #1: Type: text/plain, Size: 749 bytes --]

On Wed, Feb 08, 2012 at 11:32:10AM +0100, Christoph Egger wrote:
> On 02/08/12 02:18, Kevin O'Connor wrote:
> >We can change the seabios makefile to call these scripts with an
> >explicit $(PYTHON).
> 
> Awesome!

See the attached patch.

> >That's quite odd.  This looks data related instead of python related.
> >Try running a "make clean" and then "make" in just the seabios
> >directory.  If you still see an issue, tar up the seabios "out/"
> >directory and mail it to me.
> 
> It failed the same way.
> I sent you the tarball offlist due its size.

Looks like your version of gcc is defining sections with
.rodata.__PRETTY_FUNCTION__ - which is odd as that macro isn't used in
the code.  Does the second attached patch fix it for you?

-Kevin

[-- Attachment #2: 0001-Add-PYTHON-definition-to-Makefile.patch --]
[-- Type: text/plain, Size: 2838 bytes --]

>From 97f1ffcfac4ca382b5008a7aabfc2c300131f978 Mon Sep 17 00:00:00 2001
From: Kevin O'Connor <kevin@koconnor.net>
Date: Wed, 8 Feb 2012 20:21:29 -0500
Subject: [PATCH 1/2] Add PYTHON definition to Makefile.

Add PYTHON definition to Makefile so users can override it.  This
allows users to specify an exact python executable name to use during
the build.

Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
---
 Makefile |   11 ++++++-----
 1 files changed, 6 insertions(+), 5 deletions(-)

diff --git a/Makefile b/Makefile
index 0343ce5..513e8f1 100644
--- a/Makefile
+++ b/Makefile
@@ -63,6 +63,7 @@ export KCONFIG_CONFIG     := $(CURDIR)/.config
 OBJCOPY=objcopy
 OBJDUMP=objdump
 STRIP=strip
+PYTHON=python
 
 # Default targets
 -include $(KCONFIG_CONFIG)
@@ -152,7 +153,7 @@ $(OUT)romlayout16.lds: $(OUT)ccode32flat.o $(OUT)code32seg.o $(OUT)code16.o tool
 	$(Q)$(OBJDUMP) -thr $(OUT)code32flat.o > $(OUT)code32flat.o.objdump
 	$(Q)$(OBJDUMP) -thr $(OUT)code32seg.o > $(OUT)code32seg.o.objdump
 	$(Q)$(OBJDUMP) -thr $(OUT)code16.o > $(OUT)code16.o.objdump
-	$(Q)./tools/layoutrom.py $(OUT)code16.o.objdump $(OUT)code32seg.o.objdump $(OUT)code32flat.o.objdump $(OUT)romlayout16.lds $(OUT)romlayout32seg.lds $(OUT)romlayout32flat.lds
+	$(Q)$(PYTHON) ./tools/layoutrom.py $(OUT)code16.o.objdump $(OUT)code32seg.o.objdump $(OUT)code32flat.o.objdump $(OUT)romlayout16.lds $(OUT)romlayout32seg.lds $(OUT)romlayout32flat.lds
 
 # These are actually built by tools/layoutrom.py above, but by pulling them
 # into an extra rule we prevent make -j from spawning layoutrom.py 4 times.
@@ -174,7 +175,7 @@ $(OUT)bios.bin.elf $(OUT)bios.bin: $(OUT)rom.o tools/checkrom.py
 	@echo "  Prepping $@"
 	$(Q)$(OBJDUMP) -thr $< > $<.objdump
 	$(Q)$(OBJCOPY) -O binary $< $(OUT)bios.bin.raw
-	$(Q)./tools/checkrom.py $<.objdump $(OUT)bios.bin.raw $(OUT)bios.bin
+	$(Q)$(PYTHON) ./tools/checkrom.py $<.objdump $(OUT)bios.bin.raw $(OUT)bios.bin
 	$(Q)$(STRIP) -R .comment $< -o $(OUT)bios.bin.elf
 
 
@@ -204,15 +205,15 @@ $(OUT)vgabios.bin.raw: $(OUT)vgarom.o
 
 $(OUT)vgabios.bin: $(OUT)vgabios.bin.raw tools/buildrom.py
 	@echo "  Finalizing rom $@"
-	$(Q)./tools/buildrom.py $< $@
+	$(Q)$(PYTHON) ./tools/buildrom.py $< $@
 
 ####### dsdt build rules
 src/%.hex: src/%.dsl ./tools/acpi_extract_preprocess.py ./tools/acpi_extract.py
 	@echo "Compiling DSDT"
 	$(Q)cpp -P $< > $(OUT)$*.dsl.i.orig
-	$(Q)./tools/acpi_extract_preprocess.py $(OUT)$*.dsl.i.orig > $(OUT)$*.dsl.i
+	$(Q)$(PYTHON) ./tools/acpi_extract_preprocess.py $(OUT)$*.dsl.i.orig > $(OUT)$*.dsl.i
 	$(Q)iasl -l -tc -p $(OUT)$* $(OUT)$*.dsl.i
-	$(Q)./tools/acpi_extract.py $(OUT)$*.lst > $(OUT)$*.off
+	$(Q)$(PYTHON) ./tools/acpi_extract.py $(OUT)$*.lst > $(OUT)$*.off
 	$(Q)cat $(OUT)$*.off > $@
 
 $(OUT)ccode32flat.o: src/acpi-dsdt.hex src/ssdt-proc.hex src/ssdt-pcihp.hex
-- 
1.7.6.5


[-- Attachment #3: 0002-Permit-.rodata.__PRETTY_FUNCTION__.-sections-in-roms.patch --]
[-- Type: text/plain, Size: 1993 bytes --]

>From 805ede2bd35243a4298bc64bd81be6db7cf57f58 Mon Sep 17 00:00:00 2001
From: Kevin O'Connor <kevin@koconnor.net>
Date: Wed, 8 Feb 2012 20:23:36 -0500
Subject: [PATCH 2/2] Permit .rodata.__PRETTY_FUNCTION__. sections in roms.

Some versions of gcc appear to define these sections even though they
aren't used in the code.

Signed-off-by: Kevin O'Connor <kevin@koconnor.net>
---
 tools/layoutrom.py |   12 ++++++++----
 1 files changed, 8 insertions(+), 4 deletions(-)

diff --git a/tools/layoutrom.py b/tools/layoutrom.py
index 45738a3..86e1f33 100755
--- a/tools/layoutrom.py
+++ b/tools/layoutrom.py
@@ -147,8 +147,10 @@ def getSectionsPrefix(sections, category, prefix):
 def doLayout(sections):
     # Determine 16bit positions
     textsections = getSectionsPrefix(sections, '16', '.text.')
-    rodatasections = (getSectionsPrefix(sections, '16', '.rodata.str1.1')
-                      + getSectionsPrefix(sections, '16', '.rodata.__func__.'))
+    rodatasections = (
+        getSectionsPrefix(sections, '16', '.rodata.str1.1')
+        + getSectionsPrefix(sections, '16', '.rodata.__func__.')
+        + getSectionsPrefix(sections, '16', '.rodata.__PRETTY_FUNCTION__.'))
     datasections = getSectionsPrefix(sections, '16', '.data16.')
     fixedsections = getSectionsPrefix(sections, '16', '.fixedaddr.')
 
@@ -159,8 +161,10 @@ def doLayout(sections):
 
     # Determine 32seg positions
     textsections = getSectionsPrefix(sections, '32seg', '.text.')
-    rodatasections = (getSectionsPrefix(sections, '32seg', '.rodata.str1.1')
-                      +getSectionsPrefix(sections, '32seg', '.rodata.__func__.'))
+    rodatasections = (
+        getSectionsPrefix(sections, '32seg', '.rodata.str1.1')
+        + getSectionsPrefix(sections, '32seg', '.rodata.__func__.')
+        + getSectionsPrefix(sections, '32seg', '.rodata.__PRETTY_FUNCTION__.'))
     datasections = getSectionsPrefix(sections, '32seg', '.data32seg.')
 
     code32seg_start = setSectionsStart(
-- 
1.7.6.5


[-- Attachment #4: Type: text/plain, Size: 137 bytes --]

_______________________________________________
SeaBIOS mailing list
SeaBIOS@seabios.org
http://www.seabios.org/mailman/listinfo/seabios

Re: [SeaBIOS] seabios build failure in xen tree

[Christoph Egger]

On 02/09/12 02:28, Kevin O'Connor wrote:
> On Wed, Feb 08, 2012 at 11:32:10AM +0100, Christoph Egger wrote:
>> On 02/08/12 02:18, Kevin O'Connor wrote:
>>> We can change the seabios makefile to call these scripts with an
>>> explicit $(PYTHON).
>>
>> Awesome!
>
> See the attached patch.
>
>>> That's quite odd.  This looks data related instead of python related.
>>> Try running a "make clean" and then "make" in just the seabios
>>> directory.  If you still see an issue, tar up the seabios "out/"
>>> directory and mail it to me.
>>
>> It failed the same way.
>> I sent you the tarball offlist due its size.
>
> Looks like your version of gcc is defining sections with
> .rodata.__PRETTY_FUNCTION__ - which is odd as that macro isn't used in
> the code.  Does the second attached patch fix it for you?
>
> -Kevin

Both patches work for me and there are no follow-up errors.
Thank you.

Christoph


-- 
---to satisfy European Law for business letters:
Advanced Micro Devices GmbH
Einsteinring 24, 85689 Dornach b. Muenchen
Geschaeftsfuehrer: Alberto Bozzo, Andrew Bowd
Sitz: Dornach, Gemeinde Aschheim, Landkreis Muenchen
Registergericht Muenchen, HRB Nr. 43632