From: "H.J. Lu" <hjl.tools@gmail.com>
To: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Cc: Florian Weimer <fweimer@redhat.com>,
Adhemerval Zanella via Libc-alpha <libc-alpha@sourceware.org>,
"Carlos O'Donell" <carlos@redhat.com>,
Joseph Myers <joseph@codesourcery.com>
Subject: Re: [PATCH] elf: Replace tst-p_alignmod1-editX with a python script
Date: Wed, 26 Jan 2022 13:43:05 -0800 [thread overview]
Message-ID: <CAMe9rOr9eVK+biCzR9fGGKh6bBVF_bLNmZD2P+KG0vZfRYYdOg@mail.gmail.com> (raw)
In-Reply-To: <6fbe227d-d1cf-884e-e233-b7a80df67a58@linaro.org>
On Wed, Jan 26, 2022 at 9:11 AM Adhemerval Zanella
<adhemerval.zanella@linaro.org> wrote:
>
>
>
> On 26/01/2022 13:11, H.J. Lu wrote:
> > On Wed, Jan 26, 2022 at 7:54 AM Adhemerval Zanella
> > <adhemerval.zanella@linaro.org> wrote:
> >>
> >>
> >>
> >> On 26/01/2022 12:46, H.J. Lu wrote:
> >>> On Wed, Jan 26, 2022 at 7:45 AM Adhemerval Zanella
> >>> <adhemerval.zanella@linaro.org> wrote:
> >>>>
> >>>>
> >>>>
> >>>> On 26/01/2022 12:35, H.J. Lu wrote:
> >>>>> On Wed, Jan 26, 2022 at 4:10 AM Adhemerval Zanella
> >>>>> <adhemerval.zanella@linaro.org> wrote:
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>>> On 26/01/2022 07:09, Florian Weimer wrote:
> >>>>>>> * Adhemerval Zanella via Libc-alpha:
> >>>>>>>
> >>>>>>>> diff --git a/scripts/tst-elf-edit.py b/scripts/tst-elf-edit.py
> >>>>>>>> new file mode 100644
> >>>>>>>> index 0000000000..1fb9b8e0fe
> >>>>>>>> --- /dev/null
> >>>>>>>> +++ b/scripts/tst-elf-edit.py
> >>>>>>>> @@ -0,0 +1,206 @@
> >>>>>>>
> >>>>>>>> +class Elf_Ehdr:
> >>>>>>>> + def __init__(self, e_ident):
> >>>>>>>> + endian, addr, off = elf_types_fmts(e_ident)
> >>>>>>>> + self.fmt = '{0}HHI{1}{2}{2}IHHHHHH'.format(endian, addr, off)
> >>>>>>>> + self.len = struct.calcsize(self.fmt)
> >>>>>>>> +
> >>>>>>>> + def read(self, f):
> >>>>>>>> + buf = f.read(self.len)
> >>>>>>>> + if not buf:
> >>>>>>>> + error('{}: header too small'.format(f.name))
> >>>>>>>
> >>>>>>> I think you need to check len(buf) < self.len.
> >>>>>>
> >>>>>> Ack.
> >>>>>>
> >>>>>>>
> >>>>>>>> +class Elf_Phdr:
> >>>>>>>> + def __init__(self, e_ident):
> >>>>>>>> + endian, addr, off = elf_types_fmts(e_ident)
> >>>>>>>> + self.ei_class = e_ident[EI_CLASS]
> >>>>>>>> + if self.ei_class == ELFCLASS32:
> >>>>>>>> + self.fmt = '{0}I{2}{1}{1}IIII'.format(endian, addr, off)
> >>>>>>>> + else:
> >>>>>>>> + self.fmt = '{0}II{2}{1}{1}QQQ'.format(endian, addr, off)
> >>>>>>>> + self.len = struct.calcsize(self.fmt)
> >>>>>>>> +
> >>>>>>>> + def read(self, f):
> >>>>>>>> + buf = f.read(self.len)
> >>>>>>>> + if not buf:
> >>>>>>>> + error('{}: program header too small'.format(f.name))
> >>>>>>>
> >>>>>>> Here as well.
> >>>>>>
> >>>>>> Ack.
> >>>>>>
> >>>>>>>
> >>>>>>> Rest looks okay.
> >>>>>>
> >>>>>> Thanks, I push it upstream with the above fixes installed.
> >>>>>
> >>>>> I got
> >>>>>
> >>>>> FAIL: elf/tst-p_align3
> >>>>>
> >>>>> [hjl@gnu-tgl-2 build-x86_64-linux]$ cat elf/tst-p_align3.out
> >>>>> /export/build/gnu/tools-build/glibc-cet-gitlab/build-x86_64-linux/elf/tst-p_align3:
> >>>>> error while loading shared libraries:
> >>>>> /export/build/gnu/tools-build/glibc-cet-gitlab/build-x86_64-linux/elf/tst-p_alignmod3.so:
> >>>>> cannot change memory protections
> >>>>> [hjl@gnu-tgl-2 build-x86_64-linux]$
> >>>>>
> >>>>> Only p_align should be changed to 0x100. But I saw
> >>>>
> >>>> I am not sure if this is related because tst-p_align3 and tst-alignmod3.so uses
> >>>> LDFLAGS instead:
> >>>>
> >>>> 1644 $(objpfx)tst-align3: $(objpfx)tst-alignmod3.so
> >>>> 1645 ifeq (yes,$(have-fpie))
> >>>> 1646 CFLAGS-tst-align3.c += $(PIE-ccflag)
> >>>> 1647 endif
> >>>> 1648 LDFLAGS-tst-align3 += -Wl,-z,max-page-size=0x200000
> >>>> 1649 LDFLAGS-tst-alignmod3.so += -Wl,-z,max-page-size=0x200000
> >>>> 1650 $(objpfx)tst-alignmod3.so: $(libsupport)
> >>>>
> >>>> And the patch only changed the way tst-p_align1 and tst-p_align2 modules were
> >>>> changed.
> >>>
> >>> It is tst-p_alignmod3.so, not tst-alignmod3.so.
> >>> ^^
> >> My confusion here. But even though the script in not involved:
> >>
> >> 2622 LDFLAGS-tst-p_alignmod3.so += -Wl,-z,max-page-size=0x100,-z,common-page-size=0x100
> >> 2623
> >> 2624 $(objpfx)tst-p_align3: $(objpfx)tst-p_alignmod3.so
> >> 2625 $(objpfx)tst-p_align3.out: tst-p_align3.sh $(objpfx)tst-p_align3
> >> 2626 $(SHELL) $< $(common-objpfx) '$(test-program-prefix)'; \
> >> 2627 $(evaluate-test)
> >>
> >> That's why it is puzzling me that it start to fail now.
> >
> > This test checks the invalid p_align handling. You have to edit the
> > binary to change p_align to 0x100 which is invalid.
> >
>
> But it is already done by the LD_FLAGS above (from elf/Makefile). I am
> still failing to understand why my patch that only changes tst-p_align1
> and tst-p_align2 objects made tst-p_align3 to fail.
>
> Before:
>
> $ readelf -l elf/tst-p_align3
> [...]
> LOAD 0x0000000000000000 0x0000000000000000 0x0000000000000000
> 0x00000000000015c0 0x00000000000015c0 R 0x1000
> LOAD 0x0000000000002000 0x0000000000002000 0x0000000000002000
> 0x0000000000001892 0x0000000000001892 R E 0x1000
> LOAD 0x0000000000004000 0x0000000000004000 0x0000000000004000
> 0x0000000000000d58 0x0000000000000d58 R 0x1000
> LOAD 0x0000000000005cf8 0x0000000000006cf8 0x0000000000006cf8
> 0x0000000000000498 0x00000000000004d8 RW 0x1000
> [...]
> $ readelf -l elf/tst-p_alignmod3.so
> [....]
> Program Headers:
> Type Offset VirtAddr PhysAddr
> FileSiz MemSiz Flags Align
> LOAD 0x0000000000000000 0x0000000000000000 0x0000000000000000
> 0x0000000000000540 0x0000000000000540 R 0x100
> LOAD 0x0000000000000600 0x0000000000000600 0x0000000000000600
> 0x00000000000001f2 0x00000000000001f2 R E 0x100
> LOAD 0x0000000000000800 0x0000000000000800 0x0000000000000800
> 0x00000000000000d0 0x00000000000000d0 R 0x100
> LOAD 0x0000000000000930 0x0000000000000a30 0x0000000000000a30
>
> 0x00000000000001f8 0x0000000000000208 RW 0x100
> After:
>
> $ readelf -l elf/tst-p_align3
> [...]
> LOAD 0x0000000000000000 0x0000000000000000 0x0000000000000000
> 0x00000000000015b8 0x00000000000015b8 R 0x1000
> LOAD 0x0000000000002000 0x0000000000002000 0x0000000000002000
> 0x0000000000001892 0x0000000000001892 R E 0x1000
> LOAD 0x0000000000004000 0x0000000000004000 0x0000000000004000
> 0x0000000000000d50 0x0000000000000d50 R 0x1000
> LOAD 0x0000000000005cf8 0x0000000000006cf8 0x0000000000006cf8
> 0x0000000000000498 0x00000000000004d8 RW 0x1000
>
> $ readelf -l elf/tst-p_alignmod3.so
> Elf file type is DYN (Shared object file)
> [....]
> Program Headers:
> Type Offset VirtAddr PhysAddr
> FileSiz MemSiz Flags Align
> LOAD 0x0000000000000000 0x0000000000000000 0x0000000000000000
> 0x0000000000000540 0x0000000000000540 R 0x100
> LOAD 0x0000000000000600 0x0000000000000600 0x0000000000000600
> 0x00000000000001f2 0x00000000000001f2 R E 0x100
> LOAD 0x0000000000000800 0x0000000000000800 0x0000000000000800
> 0x00000000000000d0 0x00000000000000d0 R 0x100
> LOAD 0x0000000000000930 0x0000000000000a30 0x0000000000000a30
> 0x00000000000001f8 0x0000000000000208 RW 0x100
>
> So PT_LOAD p_align are still strictly the same.
Never mind. We need to disable GNU_RELRO segment to reliably trigger
.../elf/tst-p_alignmod3.so: ELF load command address/offset not page-aligned
A patch is posted at
https://sourceware.org/pipermail/libc-alpha/2022-January/135733.html
--
H.J.
prev parent reply other threads:[~2022-01-26 21:43 UTC|newest]
Thread overview: 10+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-01-25 17:05 Adhemerval Zanella
2022-01-26 10:09 ` Florian Weimer
2022-01-26 12:10 ` Adhemerval Zanella
2022-01-26 15:35 ` H.J. Lu
2022-01-26 15:45 ` Adhemerval Zanella
2022-01-26 15:46 ` H.J. Lu
2022-01-26 15:54 ` Adhemerval Zanella
2022-01-26 16:11 ` H.J. Lu
2022-01-26 17:11 ` Adhemerval Zanella
2022-01-26 21:43 ` H.J. Lu [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=CAMe9rOr9eVK+biCzR9fGGKh6bBVF_bLNmZD2P+KG0vZfRYYdOg@mail.gmail.com \
--to=hjl.tools@gmail.com \
--cc=adhemerval.zanella@linaro.org \
--cc=carlos@redhat.com \
--cc=fweimer@redhat.com \
--cc=joseph@codesourcery.com \
--cc=libc-alpha@sourceware.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).