Handle ARM relocations #121

frederiksdun · 2016-10-24T09:26:44Z

This PR adds support to handle ARM's R_ARM_ABS32 relocation.
Fixes issue #86

frederiksdun · 2016-10-24T10:35:20Z

Any hints why https://travis-ci.org/eliben/pyelftools/builds/170086267 failed, but 202 built?

eliben · 2016-10-24T12:21:08Z

Looking at the error log it says:

Traceback (most recent call last):

  File "scripts/readelf.py", line 1229, in <module>

    main()

  File "scripts/readelf.py", line 1189, in main

    show_heading=not do_file_header)

  File "scripts/readelf.py", line 256, in display_section_headers

    nsec, section.name, describe_sh_type(section['sh_type'])))

  File "./elftools/elf/descriptions.py", line 58, in describe_sh_type

    elif x >= ENUM_SH_TYPE['SHT_LOOS'] and x < ENUM_SH_TYPE['SHT_GNU_versym']:

TypeError: unorderable types: str() >= int()

@@ aborting - 'scripts/readelf.py' returned '1'

You could try re-running it locally on Python 3.3+ to reproduce

frederiksdun · 2016-10-24T14:12:46Z

@eliben I think the current problem is that exidx and unwind are defined as 0x70000001 which is used as key in a dict(). Is unwind a valid sh_type?

eliben · 2016-10-25T03:20:45Z

Not sure if just kicking them out is the right approach. These enum values are in "processor-specific" space which means many of them collide between processors. I think if you look in enums.py you'll see a bunch of such duplicated values, no?

frederiksdun · 2016-10-25T06:53:43Z

Yes I see duplicates in enum.py which cause problems in description.py.
Even PT_ARM_EXIDX and PT_ARM_UNWIND have the same value which would conflict if descriptions are machine aware.

Do we need the descriptions for UNWIND?

eliben · 2016-10-26T02:53:01Z

I think this is also related to #71

According to the ARM ABI it's an alias for PT_ARM_EXIDX; having them both in the same enum value confuses parsing. Keeping EXIDX rather than UNWIND because binutils seems to be using it. Related to #121

eliben · 2016-10-26T03:12:36Z

Ugh, this is a nasty issue. I think that for any given processor, all enums should be unique. I found that on ARM, EXIDX and UNWIND are synonyms (http://infocenter.arm.com/help/topic/com.arm.doc.ihi0044f/IHI0044F_aaelf.pdf) so I removed UNWIND.

However, across different processors it's valid to have duplications in that range (above 0x70000000). Here there's a deficiency in pyelftools parsing, in that it doesn't take the arch into account when parsing these fields. It should. This is also mentioned in #71.

eliben · 2016-10-26T03:13:33Z

Presumably the test file you're adding hits this. Can you try to clean it up a bit (maybe just remove these segments) so it passes all tests? Then the PR can be pulled, since the parsing issue is unrelated

frederiksdun · 2016-10-27T09:33:44Z

I rebased against master

eliben · 2016-10-28T03:11:23Z

It seems like it's still failing on Travis for Python 3.4 and 3.5 -- maybe it's still the non-determinism?

eliben · 2016-10-31T12:32:19Z

@frederiksdun I'll look into making the arch-specific field parsing work (as detailed here and in #71) -- it may unblock this PR. May take me a bit to find time for this, though

This should address #71 and #121, hopefully

eliben · 2018-01-15T22:35:31Z

A couple of recent commits made p_type and sh_type parsing dependent on e_machine, which should resolve the non-determinism issue in enums that was plaguing this.

I'll be happy to merge the PR if it's cleanly rebased on the latest master and all tests pass.

This should address eliben#71 and eliben#121, hopefully

Generated on Ubuntu 14.04 using: arm-linux-gnueabi-gcc-4.7 -c -g -o reloc_armhf_gcc.o.elf hello.c

Generated on Ubuntu 14.04 using: arm-linux-gnueabihf-gcc-4.7 -c -g -o reloc_armhf_gcc.o.elf hello.c

frederiksdun · 2018-07-16T11:06:31Z

I want to update some more details before merging

Generated on Ubuntu 14.04 using: arm-linux-gnueabihf-gcc-4.7 -g -o simple_armhf_gcc.o.elf hello.c

…m binutils 2.30

… from binutils 2.30 testcase arm-be8

frederiksdun · 2018-07-16T12:47:27Z

@eliben Feel free to merge

* relocation: handle ARM binaries * relocation: handle R_ARM_ABS32 for ARM machines * testfiles: add reloc_arm_gcc.o.elf Generated on Ubuntu 14.04 using: arm-linux-gnueabi-gcc-4.7 -c -g -o reloc_armhf_gcc.o.elf hello.c * testfiles: add reloc_armhf_gcc.o.elf Generated on Ubuntu 14.04 using: arm-linux-gnueabihf-gcc-4.7 -c -g -o reloc_armhf_gcc.o.elf hello.c * readelf: print soft-float abi for ARM if EF_ARM_ABI_FLOAT_SOFT in flags * readelf: print hard-float abi for ARM if EF_ARM_ABI_FLOAT_HARD in flags * readelf: print BE8 info for armeb binaries * testfiles: add simple_armhf_gcc.o.elf Generated on Ubuntu 14.04 using: arm-linux-gnueabihf-gcc-4.7 -g -o simple_armhf_gcc.o.elf hello.c * elf: remove unwind from dicts and set ARM_EXIDX description * testfiles: add reloc_armsf_gcc.o.elf as soft float testcase taken from binutils 2.30 * testfiles: add reloc_armeb_gcc.o.elf as arm big endian testcase taken from binutils 2.30 testcase arm-be8 * readelf: print endian info LE8 if flag was set in header flags

frederiksdun force-pushed the master branch from dee4827 to 2ae01bf Compare October 24, 2016 10:29

frederiksdun force-pushed the master branch from 2ae01bf to 3ba09e6 Compare October 24, 2016 13:14

frederiksdun force-pushed the master branch from b32c42b to 9dd0974 Compare October 24, 2016 14:38

eliben mentioned this pull request Oct 26, 2016

Non-determinism of dict enumeration results in non-deterministic parsing #71

Closed

frederiksdun force-pushed the master branch from 9dd0974 to 907d639 Compare October 27, 2016 08:11

frederiksdun force-pushed the master branch from 907d639 to e5a9ad8 Compare October 27, 2016 09:35

eliben mentioned this pull request Jan 5, 2018

Only use os- and processor- specific enum values on the appropriate os and processor #175

Closed

eliben added a commit that referenced this pull request Jan 15, 2018

e_machine-dependent p_type parsing;

7371660

This should address #71 and #121, hopefully

eliben added the pending-user-input label Jan 15, 2018

frederiksdun pushed a commit to frederiksdun/pyelftools that referenced this pull request Jul 16, 2018

e_machine-dependent p_type parsing;

9b28037

This should address eliben#71 and eliben#121, hopefully

frederiksdun added 4 commits July 16, 2018 12:28

relocation: handle ARM binaries

e2c0c5e

relocation: handle R_ARM_ABS32 for ARM machines

ed750d2

testfiles: add reloc_arm_gcc.o.elf

314a464

Generated on Ubuntu 14.04 using: arm-linux-gnueabi-gcc-4.7 -c -g -o reloc_armhf_gcc.o.elf hello.c

testfiles: add reloc_armhf_gcc.o.elf

5fda2bf

Generated on Ubuntu 14.04 using: arm-linux-gnueabihf-gcc-4.7 -c -g -o reloc_armhf_gcc.o.elf hello.c

frederiksdun force-pushed the master branch from 49513ee to 8a3215b Compare July 16, 2018 10:35

frederiksdun force-pushed the master branch from 8a3215b to c77d33f Compare July 16, 2018 11:45

frederiksdun added 8 commits July 16, 2018 14:19

readelf: print soft-float abi for ARM if EF_ARM_ABI_FLOAT_SOFT in flags

2032c29

readelf: print hard-float abi for ARM if EF_ARM_ABI_FLOAT_HARD in flags

9714076

readelf: print BE8 info for armeb binaries

9652d4b

readelf: print endian info LE8 if flag was set in header flags

cab73ea

elf: remove unwind from dicts and set ARM_EXIDX description

df7f79d

testfiles: add simple_armhf_gcc.o.elf

e193b0d

Generated on Ubuntu 14.04 using: arm-linux-gnueabihf-gcc-4.7 -g -o simple_armhf_gcc.o.elf hello.c

testfiles: add reloc_armsf_gcc.o.elf as soft float testcase taken fro…

c22d2d1

…m binutils 2.30

testfiles: add reloc_armeb_gcc.o.elf as arm big endian testcase taken…

1cf3c75

… from binutils 2.30 testcase arm-be8

frederiksdun force-pushed the master branch from c77d33f to 1cf3c75 Compare July 16, 2018 12:21

eliben approved these changes Jul 16, 2018

View reviewed changes

eliben merged commit ab76d5c into eliben:master Jul 16, 2018

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Handle ARM relocations #121

Handle ARM relocations #121

frederiksdun commented Oct 24, 2016

frederiksdun commented Oct 24, 2016

eliben commented Oct 24, 2016

frederiksdun commented Oct 24, 2016

eliben commented Oct 25, 2016

frederiksdun commented Oct 25, 2016

eliben commented Oct 26, 2016

eliben commented Oct 26, 2016

eliben commented Oct 26, 2016

frederiksdun commented Oct 27, 2016

eliben commented Oct 28, 2016

eliben commented Oct 31, 2016

eliben commented Jan 15, 2018

frederiksdun commented Jul 16, 2018

frederiksdun commented Jul 16, 2018

Handle ARM relocations #121

Handle ARM relocations #121

Conversation

frederiksdun commented Oct 24, 2016

frederiksdun commented Oct 24, 2016

eliben commented Oct 24, 2016

frederiksdun commented Oct 24, 2016

eliben commented Oct 25, 2016

frederiksdun commented Oct 25, 2016

eliben commented Oct 26, 2016

eliben commented Oct 26, 2016

eliben commented Oct 26, 2016

frederiksdun commented Oct 27, 2016

eliben commented Oct 28, 2016

eliben commented Oct 31, 2016

eliben commented Jan 15, 2018

frederiksdun commented Jul 16, 2018

frederiksdun commented Jul 16, 2018