patchelf: Assertion failed when setting interpreter

Done

Details

3 participants

Efraim Flashner
Ivan Vilata i Balaguer
Ludovic Courtès

Owner: unassigned

Submitted by: Ivan Vilata i Balaguer

Severity: normal

Ivan Vilata i Balaguer wrote on 4 Nov 2019 05:56

Recipients:(address . bug-guix@gnu.org)

Message-ID:20191104045614.GI17621@sax.terramar.selidor.net

Hi, I'm using patchelf 0.8 from Guix commit 7f81cce3 on Debian Sid. When

trying to patch the `go` binary from

https://dl.google.com/go/go1.12.3.linux-amd64.tar.gz, I get the following

error:

ivan@sax /tmp/tmps2Cv6w [env]$ patchelf --print-interpreter $SHELL

/gnu/store/ahqgl4h89xqj695lgqvsaf6zh2nhy4pj-glibc-2.29/lib/ld-linux-x86-64.so.2

ivan@sax /tmp/tmps2Cv6w [env]$ patchelf --set-interpreter $(patchelf --print-interpreter $SHELL) /tmp/tmps2Cv6w/golang/bin/go

patchelf: patchelf.cc:701: void ElfFile<Elf_Ehdr, Elf_Phdr, Elf_Shdr, Elf_Addr, Elf_Off, Elf_Dyn, Elf_Sym>::rewriteSectionsExecutable() \

[with Elf_Ehdr = Elf64_Ehdr; Elf_Phdr = Elf64_Phdr; Elf_Shdr = Elf64_Shdr; Elf_Addr = long unsigned int; Elf_Off = long unsigned int; \

Elf_Dyn = Elf64_Dyn; Elf_Sym = Elf64_Sym]: Assertion `(off_t) rdi(hdr->e_shoff) >= startOffset' failed.

Aborted

(I know Go is packed for Guix, my need arises from trying to build an

unrelated project which relies on binary Go for its build process.)

It may be the problem described here regarding Go-produced binaries:

https://github.com/NixOS/patchelf/issues/66. It seems to be fixed in

patchelf 0.10, and indeed trying the same operation with patchelf 0.10 from

Debian does succeed to patch the binary.

As an aside, I tried to build `--with-source` for 0.10 and it succeeds to

compile, but tests fail to pass.

Thank you very much!

Ivan Vilata i Balaguer -- https://elvil.net/

Ludovic Courtès wrote on 5 Nov 2019 15:12

Recipients:(name . Ivan Vilata i Balaguer)(address . ivan@selidor.net)(address . 38055@debbugs.gnu.org)

Message-ID:87muda1kfs.fsf@gnu.org

Hi Ivan,

Ivan Vilata i Balaguer <ivan@selidor.net> skribis:

Toggle quote (13 lines)> Hi, I'm using patchelf 0.8 from Guix commit 7f81cce3 on Debian Sid.  When
> trying to patch the `go` binary from
> <https://dl.google.com/go/go1.12.3.linux-amd64.tar.gz>, I get the following
> error:
>
>     ivan@sax /tmp/tmps2Cv6w [env]$ patchelf --print-interpreter $SHELL
>     /gnu/store/ahqgl4h89xqj695lgqvsaf6zh2nhy4pj-glibc-2.29/lib/ld-linux-x86-64.so.2
>     ivan@sax /tmp/tmps2Cv6w [env]$ patchelf --set-interpreter $(patchelf --print-interpreter $SHELL) /tmp/tmps2Cv6w/golang/bin/go
>     patchelf: patchelf.cc:701: void ElfFile<Elf_Ehdr, Elf_Phdr, Elf_Shdr, Elf_Addr, Elf_Off, Elf_Dyn, Elf_Sym>::rewriteSectionsExecutable() \
>      [with Elf_Ehdr = Elf64_Ehdr; Elf_Phdr = Elf64_Phdr; Elf_Shdr = Elf64_Shdr; Elf_Addr = long unsigned int; Elf_Off = long unsigned int; \
>      Elf_Dyn = Elf64_Dyn; Elf_Sym = Elf64_Sym]: Assertion `(off_t) rdi(hdr->e_shoff) >= startOffset' failed.
>     Aborted

I think it’s a bug you should report upstream to the PatchELF

maintainers; it’s probably not Guix-specific.

Thanks,

Ludo’.

Efraim Flashner wrote on 5 Nov 2019 17:18

Recipients:(name . Ludovic Courtès)(address . ludo@gnu.org)

Message-ID:20191105161822.GG14453@E5400

On Tue, Nov 05, 2019 at 03:12:23PM +0100, Ludovic Courtès wrote:

Toggle quote (21 lines)> Hi Ivan,
> 
> Ivan Vilata i Balaguer <ivan@selidor.net> skribis:
> 
> > Hi, I'm using patchelf 0.8 from Guix commit 7f81cce3 on Debian Sid.  When
> > trying to patch the `go` binary from
> > <https://dl.google.com/go/go1.12.3.linux-amd64.tar.gz>, I get the following
> > error:
> >
> >     ivan@sax /tmp/tmps2Cv6w [env]$ patchelf --print-interpreter $SHELL
> >     /gnu/store/ahqgl4h89xqj695lgqvsaf6zh2nhy4pj-glibc-2.29/lib/ld-linux-x86-64.so.2
> >     ivan@sax /tmp/tmps2Cv6w [env]$ patchelf --set-interpreter $(patchelf --print-interpreter $SHELL) /tmp/tmps2Cv6w/golang/bin/go
> >     patchelf: patchelf.cc:701: void ElfFile<Elf_Ehdr, Elf_Phdr, Elf_Shdr, Elf_Addr, Elf_Off, Elf_Dyn, Elf_Sym>::rewriteSectionsExecutable() \
> >      [with Elf_Ehdr = Elf64_Ehdr; Elf_Phdr = Elf64_Phdr; Elf_Shdr = Elf64_Shdr; Elf_Addr = long unsigned int; Elf_Off = long unsigned int; \
> >      Elf_Dyn = Elf64_Dyn; Elf_Sym = Elf64_Sym]: Assertion `(off_t) rdi(hdr->e_shoff) >= startOffset' failed.
> >     Aborted
> 
> I think it’s a bug you should report upstream to the PatchELF
> maintainers; it’s probably not Guix-specific.
> 

On the other hand, if I were the patchelf maintainers, I'd suggest
upgrading our package from 0.8 to a newer version.

-- 
Efraim Flashner   <efraim@flashner.co.il>   ????? ?????
GPG key = A28B F40C 3E55 1372 662D  14F7 41AA E7DC CA3D 8351
Confidentiality cannot be guaranteed on emails sent or received unencrypted

Ivan Vilata i Balaguer wrote on 6 Nov 2019 22:42

Recipients:(name . Efraim Flashner)(address . efraim@flashner.co.il)

Message-ID:20191106214221.GL17621@sax.terramar.selidor.net

Efraim Flashner (2019-11-05 18:18:22 +0200) wrote:

Toggle quote (23 lines)> On Tue, Nov 05, 2019 at 03:12:23PM +0100, Ludovic Courtès wrote:
> > 
> > Ivan Vilata i Balaguer <ivan@selidor.net> skribis:
> > 
> > > Hi, I'm using patchelf 0.8 from Guix commit 7f81cce3 on Debian Sid.
> > > When trying to patch the `go` binary from
> > > <https://dl.google.com/go/go1.12.3.linux-amd64.tar.gz>, I get the
> > > following error:
> > >
> > >     ivan@sax /tmp/tmps2Cv6w [env]$ patchelf --print-interpreter $SHELL
> > >     /gnu/store/ahqgl4h89xqj695lgqvsaf6zh2nhy4pj-glibc-2.29/lib/ld-linux-x86-64.so.2
> > >     ivan@sax /tmp/tmps2Cv6w [env]$ patchelf --set-interpreter $(patchelf --print-interpreter $SHELL) /tmp/tmps2Cv6w/golang/bin/go
> > >     patchelf: patchelf.cc:701: void ElfFile<Elf_Ehdr, Elf_Phdr, Elf_Shdr, Elf_Addr, Elf_Off, Elf_Dyn, Elf_Sym>::rewriteSectionsExecutable() \
> > >      [with Elf_Ehdr = Elf64_Ehdr; Elf_Phdr = Elf64_Phdr; Elf_Shdr = Elf64_Shdr; Elf_Addr = long unsigned int; Elf_Off = long unsigned int; \
> > >      Elf_Dyn = Elf64_Dyn; Elf_Sym = Elf64_Sym]: Assertion `(off_t) rdi(hdr->e_shoff) >= startOffset' failed.
> > >     Aborted
> > 
> > I think it’s a bug you should report upstream to the PatchELF
> > maintainers; it’s probably not Guix-specific.
> 
> On the other hand, if I were the patchelf maintainers, I'd suggest
> upgrading our package from 0.8 to a newer version.

Yeah, as I mentioned in the original mail that particular problem does indeed
seem to be fixed in 0.10.  However when I try to build that source with `guix
build patchelf --with-source=…`, tests fail.

If I run `guix environment -C --pure patchelf` then unpack and build the
source, the only test that actually fails is `no-rpath.sh`.  If I run `sh -x
no-rpath.sh` I get this:

```
++ basename no-rpath.sh .sh
+ SCRATCH=scratch/no-rpath
+ rm -rf scratch/no-rpath
+ mkdir -p scratch/no-rpath
+ cp no-rpath scratch/no-rpath/
++ ../src/patchelf --print-rpath scratch/no-rpath/no-rpath
+ oldRPath=/gnu/store/ahqgl4h89xqj695lgqvsaf6zh2nhy4pj-glibc-2.29/lib:/gnu/store/2plcy91lypnbbysb18ymnhaw3zwk8pg1-gcc-7.4.0-lib/lib:/gnu/store/2plcy91lypnbbysb18ymnhaw3zwk8pg1-gcc-7.4.0-lib/lib/gcc/x86_64-unknown-linux-gnu/7.4.0/../../..:/gnu/store/dcrwf5irwh39knld1wim1qkny659af9g-profile/lib
+ test -n /gnu/store/ahqgl4h89xqj695lgqvsaf6zh2nhy4pj-glibc-2.29/lib:/gnu/store/2plcy91lypnbbysb18ymnhaw3zwk8pg1-gcc-7.4.0-lib/lib:/gnu/store/2plcy91lypnbbysb18ymnhaw3zwk8pg1-gcc-7.4.0-lib/lib/gcc/x86_64-unknown-linux-gnu/7.4.0/../../..:/gnu/store/dcrwf5irwh39knld1wim1qkny659af9g-profile/lib
+ exit 1
```

To succeed, the output of `…/patchelf --print-rpath …/no-rpath`
(i.e. `oldRPath`) should be empty.  I'm not that familiar with Guix's GNU
build system, but is that at all possible under Guix?  I mean, maybe the test
is pointless or must be altered in some Guix-specific way for the `no-rpath`
binary not to have an rpath.

Cheers,

-- 
Ivan Vilata i Balaguer -- https://elvil.net/

Ludovic Courtès wrote on 7 Nov 2019 21:55

Recipients:(name . Ivan Vilata i Balaguer)(address . ivan@selidor.net)

Message-ID:87zhh7mmoo.fsf@gnu.org

Bona nit!

Ivan Vilata i Balaguer <ivan@selidor.net> skribis:

Toggle quote (26 lines)> Yeah, as I mentioned in the original mail that particular problem does indeed
> seem to be fixed in 0.10.  However when I try to build that source with `guix
> build patchelf --with-source=…`, tests fail.
>
> If I run `guix environment -C --pure patchelf` then unpack and build the
> source, the only test that actually fails is `no-rpath.sh`.  If I run `sh -x
> no-rpath.sh` I get this:
>
> ```
> ++ basename no-rpath.sh .sh
> + SCRATCH=scratch/no-rpath
> + rm -rf scratch/no-rpath
> + mkdir -p scratch/no-rpath
> + cp no-rpath scratch/no-rpath/
> ++ ../src/patchelf --print-rpath scratch/no-rpath/no-rpath
> + oldRPath=/gnu/store/ahqgl4h89xqj695lgqvsaf6zh2nhy4pj-glibc-2.29/lib:/gnu/store/2plcy91lypnbbysb18ymnhaw3zwk8pg1-gcc-7.4.0-lib/lib:/gnu/store/2plcy91lypnbbysb18ymnhaw3zwk8pg1-gcc-7.4.0-lib/lib/gcc/x86_64-unknown-linux-gnu/7.4.0/../../..:/gnu/store/dcrwf5irwh39knld1wim1qkny659af9g-profile/lib
> + test -n /gnu/store/ahqgl4h89xqj695lgqvsaf6zh2nhy4pj-glibc-2.29/lib:/gnu/store/2plcy91lypnbbysb18ymnhaw3zwk8pg1-gcc-7.4.0-lib/lib:/gnu/store/2plcy91lypnbbysb18ymnhaw3zwk8pg1-gcc-7.4.0-lib/lib/gcc/x86_64-unknown-linux-gnu/7.4.0/../../..:/gnu/store/dcrwf5irwh39knld1wim1qkny659af9g-profile/lib
> + exit 1
> ```
>
> To succeed, the output of `…/patchelf --print-rpath …/no-rpath`
> (i.e. `oldRPath`) should be empty.  I'm not that familiar with Guix's GNU
> build system, but is that at all possible under Guix?  I mean, maybe the test
> is pointless or must be altered in some Guix-specific way for the `no-rpath`
> binary not to have an rpath.

Guix’ ‘gcc’ automatically adds RUNPATH entries to glibc/lib and

gcc:lib/lib:

Toggle snippet (4 lines)

$ gcc -dumpspecs |grep -e -rpath

%{!mandroid|tno-android-ld:-L/gnu/store/ahqgl4h89xqj695lgqvsaf6zh2nhy4pj-glibc-2.29/lib %{!static:-rpath=/gnu/store/ahqgl4h89xqj695lgqvsaf6zh2nhy4pj-glibc-2.29/lib %{!static-libgcc:-rpath=/gnu/store/347y0zr1a9s2f5pkcncgi3gd0r33qq81-gcc-9.2.0-lib/lib -lgcc_s}} %{pthread:-lpthread} %{shared:-lc} %{!shared:%{profile:-lc_p}%{!profile:-lc}};:%{shared:-lc} %{!shared:%{profile:-lc_p}%{!profile:-lc}} %{!static: -ldl}}

The code that modifies GCC to do that is in (gnu packages gcc).

Thus, RUNPATH is never empty, and the test above is bound to fail.

Two possibilities: change the test to ensure that ‘--print-rpath’

returns precisely libc/lib:gcc/lib, or, if that turns out to be tricky,

skip the test (in both cases, add a comment linking to this discussion

for future reference.)

How does that sound?

Thank you!

Ludo’.

Efraim Flashner wrote on 9 Nov 2019 19:25

(no subject)

Recipients:(name . Ludovic Courtès)(address . ludo@gnu.org)

Message-ID:20191109182510.GA3954@E5400

guix-patches@gnu.org
Bcc: 
Subject: Re: bug#38055: patchelf: Assertion failed when setting interpreter
Reply-To: 
In-Reply-To: <87zhh7mmoo.fsf@gnu.org>
X-PGP-Key-ID: 0x41AAE7DCCA3D8351
X-PGP-Key: https://flashner.co.il/~efraim/efraim_flashner.asc
X-PGP-Fingerprint: A28B F40C 3E55 1372 662D  14F7 41AA E7DC CA3D 8351
I put together a patch to bump patchelf to 0.10. 'guix refresh -l
patchelf' says there's 614 dependent packages, but 'git grep ,patchelf'
and 'git grep guix\ build\ rpath' it looks like there's julia and 613
haskell packages.


-- 
Efraim Flashner   <efraim@flashner.co.il>   ????? ?????
GPG key = A28B F40C 3E55 1372 662D  14F7 41AA E7DC CA3D 8351
Confidentiality cannot be guaranteed on emails sent or received unencrypted

Attachment: 0001-gnu-patchelf-Update-to-0.10.patch

Ludovic Courtès wrote on 10 Nov 2019 15:14

Re: none

Recipients:(name . Efraim Flashner)(address . efraim@flashner.co.il)

Message-ID:87k1873jjh.fsf@gnu.org

Hi,

Efraim Flashner <efraim@flashner.co.il> skribis:

Toggle quote (14 lines)> From 2db156170a24fea36aced781faf96c839a3b7d15 Mon Sep 17 00:00:00 2001
> From: Efraim Flashner <efraim@flashner.co.il>
> Date: Sat, 9 Nov 2019 20:19:11 +0200
> Subject: [PATCH] gnu: patchelf: Update to 0.10.
>
> * gnu/packages/elf.scm (patchelf): Update to 0.10.
> [source]: Remove patches.
> [arguments]: Remove patch/rework-for-arm phase. Add phase to modify
> tests for our modified GCC package.
> [native-inputs]: Add gcc:lib.
> * gnu/packages/patches/patchelf-page-size.patch,
> * gnu/packages/patches/patchelf-rework-for-arm.patch: Remove files.
> * gnu/local.mk (dist_patch_DATA): Remove them.

[...]

Toggle quote (10 lines)> +       (modify-phases %standard-phases
> +         (add-after 'unpack 'fix-tests
> +           ;; Our GCC code ensures that RUNPATH is never empty, it includes
> +           ;; at least glibc/lib and gcc:lib/lib.
> +           (lambda* (#:key inputs #:allow-other-keys)
> +             (substitute* "tests/no-rpath.sh"
> +               (("^if test.*") "")
> +               (("/xxxxxxxxxxxxxxx") (string-append (assoc-ref inputs "gcc:lib")
> +                                                    "/lib")))

Could you complement the above comment with something like: “Thus,

disable the test that checks for an empty RUNPATH.”, or whatever is

appropriate? That will clarify the intent because it’s not obvious what

the substitution is doing if you don’t have the file at hand. :-)

Otherwise LGTM, thanks for addressing this issue!

Ludo’.

Efraim Flashner wrote on 11 Nov 2019 10:27

Recipients:(name . Ludovic Courtès)(address . ludo@gnu.org)

Message-ID:20191111092730.GF3954@E5400

Some inline comments added. Patch pushed.

-- 
Efraim Flashner   <efraim@flashner.co.il>   ????? ?????
GPG key = A28B F40C 3E55 1372 662D  14F7 41AA E7DC CA3D 8351
Confidentiality cannot be guaranteed on emails sent or received unencrypted

Closed

Ivan Vilata i Balaguer wrote on 11 Nov 2019 16:10

Recipients:(name . Efraim Flashner)(address . efraim@flashner.co.il)

Message-ID:20191111151010.GO17621@sax.terramar.selidor.net

Efraim Flashner (2019-11-11 11:27:30 +0200) wrote:

Toggle quote (2 lines)

> Some inline comments added. Patch pushed.

Wow, thank you so much for taking care of this! I did prepare a patch on

Friday but I didn't have the time to send it, plus it pales in comparison to

Efraim's and it also had the `ipfs` binary segfault after patching. I'm still

attaching it in case you're curious (of course I don't expect any of it to get

mergedï¿½`;)`).

I'll try to find a moment to test your patch and see if the `ipfs` binary

doesn't segfault, then report back.

Thanks again!

Ivan Vilata i Balaguer -- https://elvil.net/

Closed

Ivan Vilata i Balaguer wrote on 11 Nov 2019 16:30

Recipients:(name . Ivan Vilata i Balaguer)(address . ivan@selidor.net)

Message-ID:20191111153030.GP17621@sax.terramar.selidor.net

Ivan Vilata i Balaguer (2019-11-11 10:10:10 -0500) wrote:

Toggle quote (3 lines)

> […] I'm still attaching it in case you're curious (of course I don't expect

> any of it to get merged�`;)`). […]

Forgot the attachment…�`:P`

Ivan Vilata i Balaguer -- https://elvil.net/

Attachment: 0001-gnu-patchelf-Update-to-0.10.patch

Closed

Ivan Vilata i Balaguer wrote on 12 Nov 2019 06:19

Recipients:(name . Ivan Vilata i Balaguer)(address . ivan@selidor.net)

Message-ID:20191112051931.GQ17621@sax.terramar.selidor.net

Ivan Vilata i Balaguer (2019-11-11 10:10:10 -0500) wrote:

Toggle quote (7 lines)

> Efraim Flashner (2019-11-11 11:27:30 +0200) wrote:

> > Some inline comments added. Patch pushed.

> […] I'll try to find a moment to test your patch and see if the `ipfs`

> binary doesn't segfault, then report back. […]

I tried your patch, the following command in a pure container environment:

$ patchelf --set-interpreter "$(patchelf --print-interpreter /bin/sh)" /path/to/bin/go

does not trigger the assertion error (both `--set-interpreter` and

`--set-rpath` suffered from the same failure), and the resulting `go` binary

can be executed without issues.

Thank you very much for fixing this!�`:)`

Ivan Vilata i Balaguer -- https://elvil.net/

Closed

Your comment

This issue is archived.

To comment on this conversation send an email to 38055@debbugs.gnu.org

To respond to this issue using the mumi CLI, first switch to it

mumi current 38055

Then, you may apply the latest patchset in this issue (with sign off)

mumi am -- -s

Or, compose a reply to this issue

mumi compose

Or, send patches to this issue

mumi send-email *.patch

is:open	open issues
is:done	closed issues
submitter:<who>	search issue submitter
author:<who>	search by message author
date:yesterday..now	search by issue date
mdate:3m..2d	search by message date