This patch series make seabios linkable with lld.
This is beneficial for FreeBSD ports as well https://svnweb.freebsd.org/ports/head/misc/seabios/Makefile as they can drop an LLD_UNSAFE
As a maintainer of lld ELF, I have triaged numerous pieces of software. seabios is the only one making use of This drops the only instance https://sourceware.org/bugzilla/show_bug.cgi?id=12726
1, 2 and 3 are really independent and can be applied in arbitrary order. 4 depends on 3. I originally combined 3 and 4 and that was why I don't think a patch series made sense.
Fangrui Song (4): Make rom16.o linkable with lld Makefile: Change ET_EXEC to ET_REL so that lld can link bios.bin.elf romlayout32flag.lds: Use `. +=` instead of `. =` test-build.sh: Delete unneeded LD capability test
Makefile | 4 ++++ scripts/layoutrom.py | 11 ++++++++++- scripts/test-build.sh | 42 +----------------------------------------- src/romlayout.S | 2 +- 4 files changed, 16 insertions(+), 43 deletions(-)
(1) In romlayout.S, .fixedaddr.\addr sections do have not the SHF_ALLOC flag. It does not make sense to reference a SHF_ALLOC section from a non-SHF_ALLOC section via R_386_PC16. GNU ld allows it but lld will warn. Add the SHF_ALLOC flag.
(2) lld requires output section descriptions to be sorted by address. Just sort the addresses beforehand.
Signed-off-by: Fangrui Song maskray@google.com --- scripts/layoutrom.py | 4 ++++ src/romlayout.S | 2 +- 2 files changed, 5 insertions(+), 1 deletion(-)
diff --git a/scripts/layoutrom.py b/scripts/layoutrom.py index 6616721..4c55390 100755 --- a/scripts/layoutrom.py +++ b/scripts/layoutrom.py @@ -321,6 +321,10 @@ def outXRefs(sections, useseg=0, exportsyms=[], forcedelta=0):
# Write LD script includes for the given sections def outSections(sections, useseg=0): + if useseg: + sections.sort(key=lambda x: x.finalsegloc) + else: + sections.sort(key=lambda x: x.finalloc) out = "" for section in sections: loc = section.finalloc diff --git a/src/romlayout.S b/src/romlayout.S index c4a4635..a854783 100644 --- a/src/romlayout.S +++ b/src/romlayout.S @@ -587,7 +587,7 @@ entry_18:
// Specify a location in the fixed part of bios area. .macro ORG addr - .section .fixedaddr.\addr + .section .fixedaddr.\addr,"a" .endm
ORG 0xe05b
On Wed, Apr 01, 2020 at 10:29:12AM -0700, Fangrui Song wrote:
(1) In romlayout.S, .fixedaddr.\addr sections do have not the SHF_ALLOC flag. It does not make sense to reference a SHF_ALLOC section from a non-SHF_ALLOC section via R_386_PC16. GNU ld allows it but lld will warn. Add the SHF_ALLOC flag.
(2) lld requires output section descriptions to be sorted by address. Just sort the addresses beforehand.
This looks like it should be two separate patches.
I know it's a pain to redo patches, but separating out changes helps greatly when tracking down regressions via "git bisect".
Signed-off-by: Fangrui Song maskray@google.com
scripts/layoutrom.py | 4 ++++ src/romlayout.S | 2 +- 2 files changed, 5 insertions(+), 1 deletion(-)
diff --git a/scripts/layoutrom.py b/scripts/layoutrom.py index 6616721..4c55390 100755 --- a/scripts/layoutrom.py +++ b/scripts/layoutrom.py @@ -321,6 +321,10 @@ def outXRefs(sections, useseg=0, exportsyms=[], forcedelta=0):
# Write LD script includes for the given sections def outSections(sections, useseg=0):
- if useseg:
sections.sort(key=lambda x: x.finalsegloc)- else:
sections.sort(key=lambda x: x.finalloc)
This looks odd to me - there shouldn't be a need to change the sort order based on useseg, as finalloc should always have the same order as finalsegloc. Also, this code alters the input list which is confusing - perhaps use "sections = sorted(sections, key=...)".
out = "" for section in sections: loc = section.finallocdiff --git a/src/romlayout.S b/src/romlayout.S index c4a4635..a854783 100644 --- a/src/romlayout.S +++ b/src/romlayout.S @@ -587,7 +587,7 @@ entry_18:
// Specify a location in the fixed part of bios area. .macro ORG addr
.section .fixedaddr.\addr
.section .fixedaddr.\addr,"a" .endm ORG 0xe05b-- 2.26.0.rc2.310.g2932bb562d-goog _______________________________________________ SeaBIOS mailing list -- seabios@seabios.org To unsubscribe send an email to seabios-leave@seabios.org
On 2020-04-06, Kevin O'Connor wrote:
On Wed, Apr 01, 2020 at 10:29:12AM -0700, Fangrui Song wrote:
(1) In romlayout.S, .fixedaddr.\addr sections do have not the SHF_ALLOC flag. It does not make sense to reference a SHF_ALLOC section from a non-SHF_ALLOC section via R_386_PC16. GNU ld allows it but lld will warn. Add the SHF_ALLOC flag.
(2) lld requires output section descriptions to be sorted by address. Just sort the addresses beforehand.
This looks like it should be two separate patches.
I know it's a pain to redo patches, but separating out changes helps greatly when tracking down regressions via "git bisect".
I would hope [PATCH 2/4] Makefile: Change ET_EXEC to ET_REL so that lld can link bios.bin.elf can be applied first it it is acceptable.
Like I said, these 1~3 have no dependency at all. Redoing a patch series and resending as a whole may mess up the list.
Signed-off-by: Fangrui Song maskray@google.com
scripts/layoutrom.py | 4 ++++ src/romlayout.S | 2 +- 2 files changed, 5 insertions(+), 1 deletion(-)
diff --git a/scripts/layoutrom.py b/scripts/layoutrom.py index 6616721..4c55390 100755 --- a/scripts/layoutrom.py +++ b/scripts/layoutrom.py @@ -321,6 +321,10 @@ def outXRefs(sections, useseg=0, exportsyms=[], forcedelta=0):
# Write LD script includes for the given sections def outSections(sections, useseg=0):
- if useseg:
sections.sort(key=lambda x: x.finalsegloc)- else:
sections.sort(key=lambda x: x.finalloc)This looks odd to me - there shouldn't be a need to change the sort order based on useseg, as finalloc should always have the same order as finalsegloc. Also, this code alters the input list which is confusing - perhaps use "sections = sorted(sections, key=...)".
Just to confirm, I should use:
if not useseg: sections = sorted(sections, key=lambda x: x.finalloc)
out = "" for section in sections: loc = section.finallocdiff --git a/src/romlayout.S b/src/romlayout.S index c4a4635..a854783 100644 --- a/src/romlayout.S +++ b/src/romlayout.S @@ -587,7 +587,7 @@ entry_18:
// Specify a location in the fixed part of bios area. .macro ORG addr
.section .fixedaddr.\addr
.section .fixedaddr.\addr,"a" .endm ORG 0xe05b-- 2.26.0.rc2.310.g2932bb562d-goog _______________________________________________ SeaBIOS mailing list -- seabios@seabios.org To unsubscribe send an email to seabios-leave@seabios.org
On Mon, Apr 06, 2020 at 05:26:35PM -0700, Fangrui Song wrote:
On 2020-04-06, Kevin O'Connor wrote:
On Wed, Apr 01, 2020 at 10:29:12AM -0700, Fangrui Song wrote:
(1) In romlayout.S, .fixedaddr.\addr sections do have not the SHF_ALLOC flag. It does not make sense to reference a SHF_ALLOC section from a non-SHF_ALLOC section via R_386_PC16. GNU ld allows it but lld will warn. Add the SHF_ALLOC flag.
(2) lld requires output section descriptions to be sorted by address. Just sort the addresses beforehand.
This looks like it should be two separate patches.
I know it's a pain to redo patches, but separating out changes helps greatly when tracking down regressions via "git bisect".
I would hope [PATCH 2/4] Makefile: Change ET_EXEC to ET_REL so that lld can link bios.bin.elf can be applied first it it is acceptable.
Like I said, these 1~3 have no dependency at all. Redoing a patch series and resending as a whole may mess up the list.
Signed-off-by: Fangrui Song maskray@google.com
scripts/layoutrom.py | 4 ++++ src/romlayout.S | 2 +- 2 files changed, 5 insertions(+), 1 deletion(-)
diff --git a/scripts/layoutrom.py b/scripts/layoutrom.py index 6616721..4c55390 100755 --- a/scripts/layoutrom.py +++ b/scripts/layoutrom.py @@ -321,6 +321,10 @@ def outXRefs(sections, useseg=0, exportsyms=[], forcedelta=0):
# Write LD script includes for the given sections def outSections(sections, useseg=0):
- if useseg:
sections.sort(key=lambda x: x.finalsegloc)- else:
sections.sort(key=lambda x: x.finalloc)This looks odd to me - there shouldn't be a need to change the sort order based on useseg, as finalloc should always have the same order as finalsegloc. Also, this code alters the input list which is confusing - perhaps use "sections = sorted(sections, key=...)".
Just to confirm, I should use:
if not useseg: sections = sorted(sections, key=lambda x: x.finalloc)
I was proposing to unconditionally use: sections = sorted(sections, key=lambda x: x.finalloc)
-Kevin
out = "" for section in sections: loc = section.finallocdiff --git a/src/romlayout.S b/src/romlayout.S index c4a4635..a854783 100644 --- a/src/romlayout.S +++ b/src/romlayout.S @@ -587,7 +587,7 @@ entry_18:
// Specify a location in the fixed part of bios area. .macro ORG addr
.section .fixedaddr.\addr
.section .fixedaddr.\addr,"a" .endm ORG 0xe05b-- 2.26.0.rc2.310.g2932bb562d-goog _______________________________________________ SeaBIOS mailing list -- seabios@seabios.org To unsubscribe send an email to seabios-leave@seabios.org
Accepting ET_EXEC as an input file is an extremely rare GNU ld feature that lld does not intend to support, because this is error-prone.
See Linux kernel commit 90ceddcb495008ac8ba7a3dce297841efcd7d584 for another use of the dd trick.
-- Changes v1 -> v2 * Add status=none to the dd command line. This suppresses dd's stderr output. * Move dd command to a separate command
Signed-off-by: Fangrui Song maskray@google.com --- Makefile | 4 ++++ 1 file changed, 4 insertions(+)
diff --git a/Makefile b/Makefile index 5f7d537..118dec0 100644 --- a/Makefile +++ b/Makefile @@ -177,10 +177,14 @@ $(OUT)romlayout32seg.lds $(OUT)romlayout32flat.lds $(OUT)code32flat.o $(OUT)code $(OUT)rom16.o: $(OUT)code16.o $(OUT)romlayout16.lds @echo " Linking $@" $(Q)$(LD) -T $(OUT)romlayout16.lds $< -o $@ + # Change e_type to ET_REL so that it can be used to link rom.o. + # Unlike GNU ld, lld does not allow an ET_EXEC input. + printf '\1' | dd of=$@ conv=notrunc bs=1 seek=16 status=none
$(OUT)rom32seg.o: $(OUT)code32seg.o $(OUT)romlayout32seg.lds @echo " Linking $@" $(Q)$(LD) -T $(OUT)romlayout32seg.lds $< -o $@ + printf '\1' | dd of=$@ conv=notrunc bs=1 seek=16 status=none
$(OUT)rom.o: $(OUT)rom16.strip.o $(OUT)rom32seg.strip.o $(OUT)code32flat.o $(OUT)romlayout32flat.lds @echo " Linking $@"
On Wed, Apr 01, 2020 at 10:29:13AM -0700, Fangrui Song wrote:
Accepting ET_EXEC as an input file is an extremely rare GNU ld feature that lld does not intend to support, because this is error-prone.
See Linux kernel commit 90ceddcb495008ac8ba7a3dce297841efcd7d584 for another use of the dd trick.
-- Changes v1 -> v2
- Add status=none to the dd command line. This suppresses dd's stderr output.
- Move dd command to a separate command
Signed-off-by: Fangrui Song maskray@google.com
Makefile | 4 ++++ 1 file changed, 4 insertions(+)
diff --git a/Makefile b/Makefile index 5f7d537..118dec0 100644 --- a/Makefile +++ b/Makefile @@ -177,10 +177,14 @@ $(OUT)romlayout32seg.lds $(OUT)romlayout32flat.lds $(OUT)code32flat.o $(OUT)code $(OUT)rom16.o: $(OUT)code16.o $(OUT)romlayout16.lds @echo " Linking $@" $(Q)$(LD) -T $(OUT)romlayout16.lds $< -o $@
- # Change e_type to ET_REL so that it can be used to link rom.o.
- # Unlike GNU ld, lld does not allow an ET_EXEC input.
- printf '\1' | dd of=$@ conv=notrunc bs=1 seek=16 status=none
$(OUT)rom32seg.o: $(OUT)code32seg.o $(OUT)romlayout32seg.lds @echo " Linking $@" $(Q)$(LD) -T $(OUT)romlayout32seg.lds $< -o $@
- printf '\1' | dd of=$@ conv=notrunc bs=1 seek=16 status=none
My high-level feedback is that the above is very fragile. I'd be reluctant to adopt that hack. What is the underlying issue that needs to be addressed?
-Kevin
$(OUT)rom.o: $(OUT)rom16.strip.o $(OUT)rom32seg.strip.o $(OUT)code32flat.o $(OUT)romlayout32flat.lds @echo " Linking $@" -- 2.26.0.rc2.310.g2932bb562d-goog _______________________________________________ SeaBIOS mailing list -- seabios@seabios.org To unsubscribe send an email to seabios-leave@seabios.org
On 2020-04-06, Kevin O'Connor wrote:
On Wed, Apr 01, 2020 at 10:29:13AM -0700, Fangrui Song wrote:
Accepting ET_EXEC as an input file is an extremely rare GNU ld feature that lld does not intend to support, because this is error-prone.
See Linux kernel commit 90ceddcb495008ac8ba7a3dce297841efcd7d584 for another use of the dd trick.
-- Changes v1 -> v2
- Add status=none to the dd command line. This suppresses dd's stderr output.
- Move dd command to a separate command
Signed-off-by: Fangrui Song maskray@google.com
Makefile | 4 ++++ 1 file changed, 4 insertions(+)
diff --git a/Makefile b/Makefile index 5f7d537..118dec0 100644 --- a/Makefile +++ b/Makefile @@ -177,10 +177,14 @@ $(OUT)romlayout32seg.lds $(OUT)romlayout32flat.lds $(OUT)code32flat.o $(OUT)code $(OUT)rom16.o: $(OUT)code16.o $(OUT)romlayout16.lds @echo " Linking $@" $(Q)$(LD) -T $(OUT)romlayout16.lds $< -o $@
- # Change e_type to ET_REL so that it can be used to link rom.o.
- # Unlike GNU ld, lld does not allow an ET_EXEC input.
- printf '\1' | dd of=$@ conv=notrunc bs=1 seek=16 status=none
$(OUT)rom32seg.o: $(OUT)code32seg.o $(OUT)romlayout32seg.lds @echo " Linking $@" $(Q)$(LD) -T $(OUT)romlayout32seg.lds $< -o $@
- printf '\1' | dd of=$@ conv=notrunc bs=1 seek=16 status=none
My high-level feedback is that the above is very fragile. I'd be reluctant to adopt that hack. What is the underlying issue that needs to be addressed?
-Kevin
lld does not take ET_EXEC as input. This is a deliberate choice. GNU gold does not accept ET_EXEC as well:
% gold a gold: error: a: unsupported ELF file type 2
1 ET_REL represents object files (.o) 2 ET_EXEC represents position-dependent executables. 3 ET_DYN represents shared objects (.so) (or PIE; for linking purposes, PIE cannot be accepted) 4 ET_CORE represents core files. For linking purposes, they cannot be accepted.
I don't know how GNU ld ends up accepting ET_EXEC. I am not even sure it is an intentional decision. A lot of sections will not be meaningful to the linker and accidentally mixing an ET_EXEC can likely lead to hard-to-debug linking issues.
I made a similar change to Linux recently. https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?i...
$(OUT)rom.o: $(OUT)rom16.strip.o $(OUT)rom32seg.strip.o $(OUT)code32flat.o $(OUT)romlayout32flat.lds @echo " Linking $@" -- 2.26.0.rc2.310.g2932bb562d-goog _______________________________________________ SeaBIOS mailing list -- seabios@seabios.org To unsubscribe send an email to seabios-leave@seabios.org
This improves the portability of the linker script and allows lld to link rom.o
Dot assignment inside an output section has an inconsistent behavior which makes lld difficult to implement. See https://bugs.llvm.org/show_bug.cgi?id=43083
Dropping `. =` turns out to be beneficial to older GNU ld as well because we can delete an ld check detecting "cannot move location counter backwards".
Signed-off-by: Fangrui Song maskray@google.com --- scripts/layoutrom.py | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-)
diff --git a/scripts/layoutrom.py b/scripts/layoutrom.py index 4c55390..94d4412 100755 --- a/scripts/layoutrom.py +++ b/scripts/layoutrom.py @@ -339,14 +339,19 @@ def outRelSections(sections, startsym, useseg=0): if section.finalloc is not None] sections.sort(key=operator.itemgetter(0)) out = "" + location = "_reloc_init_end" for addr, section in sections: loc = section.finalloc if useseg: loc = section.finalsegloc - out += ". = ( 0x%x - %s ) ;\n" % (loc, startsym) + if location == "_reloc_init_end": + out += ". += 0x%x - %s ;\n" % (loc, location) + elif location < loc: + out += ". += 0x%x ;\n" % (loc-location,) if section.name in ('.rodata.str1.1', '.rodata'): out += "_rodata%s = . ;\n" % (section.fileid,) out += "*%s.*(%s)\n" % (section.fileid, section.name) + location = loc + section.size return out
# Build linker script output for a list of relocations.
On Wed, Apr 01, 2020 at 10:29:14AM -0700, Fangrui Song wrote:
This improves the portability of the linker script and allows lld to link rom.o
Dot assignment inside an output section has an inconsistent behavior which makes lld difficult to implement. See https://bugs.llvm.org/show_bug.cgi?id=43083
Dropping `. =` turns out to be beneficial to older GNU ld as well because we can delete an ld check detecting "cannot move location counter backwards".
Signed-off-by: Fangrui Song maskray@google.com
scripts/layoutrom.py | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-)
diff --git a/scripts/layoutrom.py b/scripts/layoutrom.py index 4c55390..94d4412 100755 --- a/scripts/layoutrom.py +++ b/scripts/layoutrom.py @@ -339,14 +339,19 @@ def outRelSections(sections, startsym, useseg=0): if section.finalloc is not None] sections.sort(key=operator.itemgetter(0)) out = ""
- location = "_reloc_init_end" for addr, section in sections: loc = section.finalloc if useseg: loc = section.finalsegloc
out += ". = ( 0x%x - %s ) ;\n" % (loc, startsym)
if location == "_reloc_init_end":out += ". += 0x%x - %s ;\n" % (loc, location)elif location < loc:out += ". += 0x%x ;\n" % (loc-location,) if section.name in ('.rodata.str1.1', '.rodata'): out += "_rodata%s = . ;\n" % (section.fileid,) out += "*%s.*(%s)\n" % (section.fileid, section.name)location = loc + section.size
I'm finding this code confusing. I would recommend "location" always be a string or always be an integer - mixing them makes the code hard to follow. Also, this code removes the only reference to startsym, but doesn't update the function signature. Similarly, it hardcodes "_reloc_init_end" when I think that should be passed to the function.
-Kevin
return out# Build linker script output for a list of relocations.
2.26.0.rc2.310.g2932bb562d-goog _______________________________________________ SeaBIOS mailing list -- seabios@seabios.org To unsubscribe send an email to seabios-leave@seabios.org
On 2020-04-06, Kevin O'Connor wrote:
On Wed, Apr 01, 2020 at 10:29:14AM -0700, Fangrui Song wrote:
This improves the portability of the linker script and allows lld to link rom.o
Dot assignment inside an output section has an inconsistent behavior which makes lld difficult to implement. See https://bugs.llvm.org/show_bug.cgi?id=43083
Dropping `. =` turns out to be beneficial to older GNU ld as well because we can delete an ld check detecting "cannot move location counter backwards".
Signed-off-by: Fangrui Song maskray@google.com
scripts/layoutrom.py | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-)
diff --git a/scripts/layoutrom.py b/scripts/layoutrom.py index 4c55390..94d4412 100755 --- a/scripts/layoutrom.py +++ b/scripts/layoutrom.py @@ -339,14 +339,19 @@ def outRelSections(sections, startsym, useseg=0): if section.finalloc is not None] sections.sort(key=operator.itemgetter(0)) out = ""
- location = "_reloc_init_end" for addr, section in sections: loc = section.finalloc if useseg: loc = section.finalsegloc
out += ". = ( 0x%x - %s ) ;\n" % (loc, startsym)
if location == "_reloc_init_end":out += ". += 0x%x - %s ;\n" % (loc, location)elif location < loc:out += ". += 0x%x ;\n" % (loc-location,) if section.name in ('.rodata.str1.1', '.rodata'): out += "_rodata%s = . ;\n" % (section.fileid,) out += "*%s.*(%s)\n" % (section.fileid, section.name)location = loc + section.sizeI'm finding this code confusing. I would recommend "location" always be a string or always be an integer - mixing them makes the code hard to follow. Also, this code removes the only reference to startsym, but doesn't update the function signature. Similarly, it hardcodes "_reloc_init_end" when I think that should be passed to the function.
-Kevin
Thanks. Addressed in [PATCH v3 4/5] romlayout32flag.lds: Use `. +=` instead of `. =`
Non-zero useseg does not appear to be used, but I don't understand it enough to fix it.
return out# Build linker script output for a list of relocations.
2.26.0.rc2.310.g2932bb562d-goog _______________________________________________ SeaBIOS mailing list -- seabios@seabios.org To unsubscribe send an email to seabios-leave@seabios.org
The previous commit changed romlayout32flag.lds to use `. += ` instead of `. =`. We no longer need the LD capability test checking https://sourceware.org/bugzilla/show_bug.cgi?id=12726
Signed-off-by: Fangrui Song maskray@google.com --- scripts/test-build.sh | 42 +----------------------------------------- 1 file changed, 1 insertion(+), 41 deletions(-)
diff --git a/scripts/test-build.sh b/scripts/test-build.sh index 25cc2f2..8b35d6f 100755 --- a/scripts/test-build.sh +++ b/scripts/test-build.sh @@ -4,50 +4,10 @@ mkdir -p ${OUT} TMPFILE1=${OUT}/tmp_testcompile1.c TMPFILE1o=${OUT}/tmp_testcompile1.o -TMPFILE1_ld=${OUT}/tmp_testcompile1.lds TMPFILE2=${OUT}/tmp_testcompile2.c TMPFILE2o=${OUT}/tmp_testcompile2.o TMPFILE3o=${OUT}/tmp_testcompile3.o
-# Test if ld's alignment handling is correct. This is a known problem -# with the linker that ships with Ubuntu 11.04. -cat - > $TMPFILE1 <<EOF -const char v1[] __attribute__((section(".text.v1"))) = "0123456789"; -const char v2[] __attribute__((section(".text.v2"))) = "0123456789"; -EOF -cat - > $TMPFILE1_ld <<EOF -SECTIONS -{ - .mysection 0x88f0 : { -. = 0x10 ; -*(.text.v1) -. = 0x20 ; -*(.text.v2) -. = 0x30 ; - } -} -EOF -$CC -O -g -c $TMPFILE1 -o $TMPFILE1o > /dev/null 2>&1 -if [ $? -ne 0 ]; then - echo "Unable to execute the C compiler ($CC)." >&2 - echo "" >&2 - echo "Please install a working compiler and retry." >&2 - echo -1 - exit 0 -fi -$LD -T $TMPFILE1_ld $TMPFILE1o -o $TMPFILE2o > /dev/null 2>&1 -if [ $? -ne 0 ]; then - echo "The version of LD on this system ($LD) does not properly handle" >&2 - echo "alignments. As a result, this project can not be built." >&2 - echo "" >&2 - echo "The problem may be the result of this LD bug report:" >&2 - echo " http://sourceware.org/bugzilla/show_bug.cgi?id=12726" >&2 - echo "" >&2 - echo "Please update to a working version of binutils and retry." >&2 - echo -1 - exit 0 -fi - # Test for "-fwhole-program". Older versions of gcc (pre v4.1) don't # support the whole-program optimization - detect that. $CC -fwhole-program -S -o /dev/null -xc /dev/null > /dev/null 2>&1 @@ -87,4 +47,4 @@ echo 0 # "ebp" register is clobberred in an "asm" statement. The code has # been modified to not clobber "ebp" - no test is available yet.
-rm -f $TMPFILE1 $TMPFILE1o $TMPFILE1_ld $TMPFILE2 $TMPFILE2o $TMPFILE3o +rm -f $TMPFILE1 $TMPFILE1o $TMPFILE2 $TMPFILE2o $TMPFILE3o
On Wed, Apr 01, 2020 at 10:29:15AM -0700, Fangrui Song wrote:
The previous commit changed romlayout32flag.lds to use `. += ` instead of `. =`. We no longer need the LD capability test checking https://sourceware.org/bugzilla/show_bug.cgi?id=12726
Thanks. It would be great to remove this test. Were you able to bring up that old version of Ubuntu and verify that it produces bootable code (or definitely errors) with the new code though? I'd like to make sure the new code doesn't result in silently broken code.
-Kevin
Signed-off-by: Fangrui Song maskray@google.com
scripts/test-build.sh | 42 +----------------------------------------- 1 file changed, 1 insertion(+), 41 deletions(-)
diff --git a/scripts/test-build.sh b/scripts/test-build.sh index 25cc2f2..8b35d6f 100755 --- a/scripts/test-build.sh +++ b/scripts/test-build.sh @@ -4,50 +4,10 @@ mkdir -p ${OUT} TMPFILE1=${OUT}/tmp_testcompile1.c TMPFILE1o=${OUT}/tmp_testcompile1.o -TMPFILE1_ld=${OUT}/tmp_testcompile1.lds TMPFILE2=${OUT}/tmp_testcompile2.c TMPFILE2o=${OUT}/tmp_testcompile2.o TMPFILE3o=${OUT}/tmp_testcompile3.o
-# Test if ld's alignment handling is correct. This is a known problem -# with the linker that ships with Ubuntu 11.04. -cat - > $TMPFILE1 <<EOF -const char v1[] __attribute__((section(".text.v1"))) = "0123456789"; -const char v2[] __attribute__((section(".text.v2"))) = "0123456789"; -EOF -cat - > $TMPFILE1_ld <<EOF -SECTIONS -{
.mysection 0x88f0 : {-. = 0x10 ; -*(.text.v1) -. = 0x20 ; -*(.text.v2) -. = 0x30 ;
}-} -EOF -$CC -O -g -c $TMPFILE1 -o $TMPFILE1o > /dev/null 2>&1 -if [ $? -ne 0 ]; then
- echo "Unable to execute the C compiler ($CC)." >&2
- echo "" >&2
- echo "Please install a working compiler and retry." >&2
- echo -1
- exit 0
-fi -$LD -T $TMPFILE1_ld $TMPFILE1o -o $TMPFILE2o > /dev/null 2>&1 -if [ $? -ne 0 ]; then
- echo "The version of LD on this system ($LD) does not properly handle" >&2
- echo "alignments. As a result, this project can not be built." >&2
- echo "" >&2
- echo "The problem may be the result of this LD bug report:" >&2
- echo " http://sourceware.org/bugzilla/show_bug.cgi?id=12726" >&2
- echo "" >&2
- echo "Please update to a working version of binutils and retry." >&2
- echo -1
- exit 0
-fi
# Test for "-fwhole-program". Older versions of gcc (pre v4.1) don't # support the whole-program optimization - detect that. $CC -fwhole-program -S -o /dev/null -xc /dev/null > /dev/null 2>&1 @@ -87,4 +47,4 @@ echo 0 # "ebp" register is clobberred in an "asm" statement. The code has # been modified to not clobber "ebp" - no test is available yet.
-rm -f $TMPFILE1 $TMPFILE1o $TMPFILE1_ld $TMPFILE2 $TMPFILE2o $TMPFILE3o
+rm -f $TMPFILE1 $TMPFILE1o $TMPFILE2 $TMPFILE2o $TMPFILE3o
2.26.0.rc2.310.g2932bb562d-goog _______________________________________________ SeaBIOS mailing list -- seabios@seabios.org To unsubscribe send an email to seabios-leave@seabios.org
On 2020-04-06, Kevin O'Connor wrote:
On Wed, Apr 01, 2020 at 10:29:15AM -0700, Fangrui Song wrote:
The previous commit changed romlayout32flag.lds to use `. += ` instead of `. =`. We no longer need the LD capability test checking https://sourceware.org/bugzilla/show_bug.cgi?id=12726
Thanks. It would be great to remove this test. Were you able to bring up that old version of Ubuntu and verify that it produces bootable code (or definitely errors) with the new code though? I'd like to make sure the new code doesn't result in silently broken code.
-Kevin
I cannot test it easily. Can you do that for me, please?
Signed-off-by: Fangrui Song maskray@google.com
scripts/test-build.sh | 42 +----------------------------------------- 1 file changed, 1 insertion(+), 41 deletions(-)
diff --git a/scripts/test-build.sh b/scripts/test-build.sh index 25cc2f2..8b35d6f 100755 --- a/scripts/test-build.sh +++ b/scripts/test-build.sh @@ -4,50 +4,10 @@ mkdir -p ${OUT} TMPFILE1=${OUT}/tmp_testcompile1.c TMPFILE1o=${OUT}/tmp_testcompile1.o -TMPFILE1_ld=${OUT}/tmp_testcompile1.lds TMPFILE2=${OUT}/tmp_testcompile2.c TMPFILE2o=${OUT}/tmp_testcompile2.o TMPFILE3o=${OUT}/tmp_testcompile3.o
-# Test if ld's alignment handling is correct. This is a known problem -# with the linker that ships with Ubuntu 11.04. -cat - > $TMPFILE1 <<EOF -const char v1[] __attribute__((section(".text.v1"))) = "0123456789"; -const char v2[] __attribute__((section(".text.v2"))) = "0123456789"; -EOF -cat - > $TMPFILE1_ld <<EOF -SECTIONS -{
.mysection 0x88f0 : {-. = 0x10 ; -*(.text.v1) -. = 0x20 ; -*(.text.v2) -. = 0x30 ;
}-} -EOF -$CC -O -g -c $TMPFILE1 -o $TMPFILE1o > /dev/null 2>&1 -if [ $? -ne 0 ]; then
- echo "Unable to execute the C compiler ($CC)." >&2
- echo "" >&2
- echo "Please install a working compiler and retry." >&2
- echo -1
- exit 0
-fi -$LD -T $TMPFILE1_ld $TMPFILE1o -o $TMPFILE2o > /dev/null 2>&1 -if [ $? -ne 0 ]; then
- echo "The version of LD on this system ($LD) does not properly handle" >&2
- echo "alignments. As a result, this project can not be built." >&2
- echo "" >&2
- echo "The problem may be the result of this LD bug report:" >&2
- echo " http://sourceware.org/bugzilla/show_bug.cgi?id=12726" >&2
- echo "" >&2
- echo "Please update to a working version of binutils and retry." >&2
- echo -1
- exit 0
-fi
# Test for "-fwhole-program". Older versions of gcc (pre v4.1) don't # support the whole-program optimization - detect that. $CC -fwhole-program -S -o /dev/null -xc /dev/null > /dev/null 2>&1 @@ -87,4 +47,4 @@ echo 0 # "ebp" register is clobberred in an "asm" statement. The code has # been modified to not clobber "ebp" - no test is available yet.
-rm -f $TMPFILE1 $TMPFILE1o $TMPFILE1_ld $TMPFILE2 $TMPFILE2o $TMPFILE3o
+rm -f $TMPFILE1 $TMPFILE1o $TMPFILE2 $TMPFILE2o $TMPFILE3o
2.26.0.rc2.310.g2932bb562d-goog _______________________________________________ SeaBIOS mailing list -- seabios@seabios.org To unsubscribe send an email to seabios-leave@seabios.org
-
Fangrui Song -
Kevin O'Connor