public inbox for libc-alpha@sourceware.org
 help / color / mirror / Atom feed
From: "H.J. Lu" <hjl.tools@gmail.com>
To: Florian Weimer <fweimer@redhat.com>
Cc: GNU C Library <libc-alpha@sourceware.org>
Subject: Re: [PATCH 1/5] Linux: Implement a useful version of _startup_fatal
Date: Tue, 3 May 2022 08:54:41 -0700	[thread overview]
Message-ID: <CAMe9rOo3iPYVT0F2mGkrSht13KDt_Hm9TXD+8EQSe0L7gdrwBw@mail.gmail.com> (raw)
In-Reply-To: <87bkwfxkhg.fsf@oldenburg.str.redhat.com>

On Mon, May 2, 2022 at 10:18 PM Florian Weimer <fweimer@redhat.com> wrote:
>
> * Florian Weimer:
>
> > * H. J. Lu:
> >
> >> On Mon, May 2, 2022 at 12:53 PM Florian Weimer <fweimer@redhat.com> wrote:
> >>>
> >>> * H. J. Lu:
> >>>
> >>> >> +#include_next <startup.h>
> >>> >> diff --git a/sysdeps/unix/sysv/linux/startup.h b/sysdeps/unix/sysv/linux/startup.h
> >>> >> new file mode 100644
> >>> >> index 0000000000..a5de941759
> >>> >> --- /dev/null
> >>> >> +++ b/sysdeps/unix/sysv/linux/startup.h
> >>> >> @@ -0,0 +1,38 @@
> >>> >> +/* Linux definitions of functions used by static libc main startup.
> >>> >> +   Copyright (C) 2017-2022 Free Software Foundation, Inc.
> >>> >                              Shouldn't it just be 2022?
> >>> >> +   This file is part of the GNU C Library.
> >>> >> +
> >>> >> +   The GNU C Library is free software; you can redistribute it and/or
> >>> >> +   modify it under the terms of the GNU Lesser General Public
> >>> >> +   License as published by the Free Software Foundation; either
> >>> >> +   version 2.1 of the License, or (at your option) any later version.
> >>> >> +
> >>> >> +   The GNU C Library is distributed in the hope that it will be useful,
> >>> >> +   but WITHOUT ANY WARRANTY; without even the implied warranty of
> >>> >> +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> >>> >> +   Lesser General Public License for more details.
> >>> >> +
> >>> >> +   You should have received a copy of the GNU Lesser General Public
> >>> >> +   License along with the GNU C Library; if not, see
> >>> >> +   <https://www.gnu.org/licenses/>.  */
> >>> >> +
> >>> >> +#ifdef SHARED
> >>> >> +# include_next <startup.h>
> >>> >> +#else
> >>> >> +# include <sysdep.h>
> >>> >> +
> >>> >> +/* Avoid a run-time invocation of strlen.  */
> >>> >> +#define _startup_fatal(message)                                         \
> >>> >
> >>> > Is this needed only for static PIE?
> >>>
> >>> I don't know.  It's used from csu/libc-tls.c, some targets may need it
> >>> for ET_EXEC static linking, too.  I haven't checked.  The additional
> >>> code size is minimal.
> >>
> >> _startup_fatal was added for i386 static PIE.  I don't think it is
> >> needed for normal static executables.
> >
> > Based on the current initialization ordering, it seems to me that we
> > need it for non-PIE static programs, too.  Maybe I should check this
> > tomorrow.
>
> It does not work on x86-64 because cancellation handling needs the TCB:
>
> Breakpoint 1, _dl_early_allocate (size=size@entry=3520) at ../sysdeps/unix/sysv/linux/dl-early_allocate.c:41
> 41        if (__curbrk != NULL)
> (gdb) finish
> Run till exit from #0  _dl_early_allocate (size=size@entry=3520)
>     at ../sysdeps/unix/sysv/linux/dl-early_allocate.c:41
> __libc_setup_tls () at libc-tls.c:149
> 149       if (tlsblock == NULL)
> Value returned is $1 = (void *) 0x4eb000
> (gdb) set $rax = 0
> (gdb) c
> Continuing.
>
> Program received signal SIGSEGV, Segmentation fault.
> __writev (fd=fd@entry=2, iov=iov@entry=0x7fffffffddb0, iovcnt=iovcnt@entry=1) at ../sysdeps/unix/sysv/linux/writev.c:26
> 26        return SYSCALL_CANCEL (writev, fd, iov, iovcnt);
> (gdb) disas
> Dump of assembler code for function __writev:
> => 0x000000000044e3c0 <+0>:     mov    %fs:0x18,%eax
>    0x000000000044e3c8 <+8>:     test   %eax,%eax
>    0x000000000044e3ca <+10>:    jne    0x44e3e0 <__writev+32>
>    0x000000000044e3cc <+12>:    mov    $0x14,%eax
>    0x000000000044e3d1 <+17>:    syscall
>    0x000000000044e3d3 <+19>:    cmp    $0xfffffffffffff000,%rax
>    0x000000000044e3d9 <+25>:    ja     0x44e430 <__writev+112>
>    0x000000000044e3db <+27>:    ret
>    0x000000000044e3dc <+28>:    nopl   0x0(%rax)
>    0x000000000044e3e0 <+32>:    sub    $0x28,%rsp
>    0x000000000044e3e4 <+36>:    mov    %edx,0x1c(%rsp)
>    0x000000000044e3e8 <+40>:    mov    %rsi,0x10(%rsp)
>    0x000000000044e3ed <+45>:    mov    %edi,0x8(%rsp)
>    0x000000000044e3f1 <+49>:    call   0x46d2b0 <__pthread_enable_asynccancel>
>    0x000000000044e3f6 <+54>:    mov    0x1c(%rsp),%edx
>    0x000000000044e3fa <+58>:    mov    0x10(%rsp),%rsi
>    0x000000000044e3ff <+63>:    mov    %eax,%r8d
>    0x000000000044e402 <+66>:    mov    0x8(%rsp),%edi
>    0x000000000044e406 <+70>:    mov    $0x14,%eax
>    0x000000000044e40b <+75>:    syscall
>    0x000000000044e40d <+77>:    cmp    $0xfffffffffffff000,%rax
>    0x000000000044e413 <+83>:    ja     0x44e448 <__writev+136>
>    0x000000000044e415 <+85>:    mov    %r8d,%edi
>    0x000000000044e418 <+88>:    mov    %rax,0x8(%rsp)
>    0x000000000044e41d <+93>:    call   0x46d330 <__pthread_disable_asynccancel>
>    0x000000000044e422 <+98>:    mov    0x8(%rsp),%rax
>    0x000000000044e427 <+103>:   add    $0x28,%rsp
>    0x000000000044e42b <+107>:   ret
>    0x000000000044e42c <+108>:   nopl   0x0(%rax)
>    0x000000000044e430 <+112>:   mov    $0xffffffffffffffa8,%rdx
>    0x000000000044e437 <+119>:   neg    %eax
>    0x000000000044e439 <+121>:   mov    %eax,%fs:(%rdx)
>    0x000000000044e43c <+124>:   mov    $0xffffffffffffffff,%rax
>    0x000000000044e443 <+131>:   ret
>    0x000000000044e444 <+132>:   nopl   0x0(%rax)
>    0x000000000044e448 <+136>:   mov    $0xffffffffffffffa8,%rdx
>    0x000000000044e44f <+143>:   neg    %eax
>    0x000000000044e451 <+145>:   mov    %eax,%fs:(%rdx)
>    0x000000000044e454 <+148>:   mov    $0xffffffffffffffff,%rax
>    0x000000000044e45b <+155>:   jmp    0x44e415 <__writev+85>
> End of assembler dump.
> (gdb) bt
> #0  __writev (fd=fd@entry=2, iov=iov@entry=0x7fffffffddb0,
>     iovcnt=iovcnt@entry=1) at ../sysdeps/unix/sysv/linux/writev.c:26
> #1  0x00000000004077aa in writev_for_fatal (fd=<optimized out>,
>     total=<optimized out>, niov=<optimized out>, iov=<optimized out>)
>     at ../sysdeps/posix/libc_fatal.c:44
> #2  __libc_message (action=action@entry=do_abort, fmt=fmt@entry=0x4c70ad "%s")
>     at ../sysdeps/posix/libc_fatal.c:124
> #3  0x000000000040789c in __libc_fatal (
>     message=message@entry=0x4adb60 "Fatal glibc error: Cannot allocate TLS block\n") at ../sysdeps/posix/libc_fatal.c:164
> #4  0x000000000040a3c0 in __libc_setup_tls () at libc-tls.c:150
> #5  0x0000000000409709 in __libc_start_main_impl (main=0x401740 <main>,
>     argc=1, argv=0x7fffffffe008, init=<optimized out>, fini=<optimized out>,
>     rtld_fini=0x0, stack_end=0x7fffffffdff8) at ../csu/libc-start.c:301
> #6  0x0000000000401da1 in _start () at ../sysdeps/x86_64/start.S:115
> (gdb)
>
> That might be a different bug.

The startup codes can't use cancellation before it is ready.  Do we need to
check SHARED in this case?

-- 
H.J.

  reply	other threads:[~2022-05-03 15:55 UTC|newest]

Thread overview: 23+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-05-02 19:15 [PATCH 0/5] Linux: Fall back to mmap if early sbrk fails Florian Weimer
2022-05-02 19:15 ` [PATCH 1/5] Linux: Implement a useful version of _startup_fatal Florian Weimer
2022-05-02 19:45   ` H.J. Lu
2022-05-02 19:52     ` Florian Weimer
2022-05-02 20:06       ` H.J. Lu
2022-05-02 20:37         ` Florian Weimer
2022-05-03  5:18           ` Florian Weimer
2022-05-03 15:54             ` H.J. Lu [this message]
2022-05-03 16:00               ` Florian Weimer
2022-05-03 16:36                 ` H.J. Lu
2022-05-03 16:43                   ` Florian Weimer
2022-05-03 16:46                     ` H.J. Lu
2022-05-03 16:57                       ` Florian Weimer
2022-05-03 17:01                         ` Adhemerval Zanella
2022-05-03 20:16                           ` Florian Weimer
2022-05-02 19:15 ` [PATCH 2/5] Linux: Define MMAP_CALL_INTERNAL Florian Weimer
2022-05-03 18:30   ` Florian Weimer
2022-05-04 13:10     ` Stefan Liebler
2022-05-02 19:15 ` [PATCH 3/5] i386: Remove OPTIMIZE_FOR_GCC_5 from Linux libc-do-syscall.S Florian Weimer
2022-05-02 19:20   ` H.J. Lu
2022-05-02 19:15 ` [PATCH 4/5] i386: Honor I386_USE_SYSENTER for 6-argument Linux system calls Florian Weimer
2022-05-02 19:24   ` H.J. Lu
2022-05-02 19:15 ` [PATCH 5/5] csu: Implement and use _dl_early_allocate during static startup Florian Weimer

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=CAMe9rOo3iPYVT0F2mGkrSht13KDt_Hm9TXD+8EQSe0L7gdrwBw@mail.gmail.com \
    --to=hjl.tools@gmail.com \
    --cc=fweimer@redhat.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).