mbox series

[for-8.1,v10,00/14] linux-user: image mapping fixes

Message ID 20230807163705.9848-1-richard.henderson@linaro.org
Headers show
Series linux-user: image mapping fixes | expand

Message

Richard Henderson Aug. 7, 2023, 4:36 p.m. UTC
This is the second half of

https://patchew.org/QEMU/20230804220032.295411-1-richard.henderson@linaro.org/

which I held back because of regressions with s390x testing.

It turns out that patch 4, "Use MAP_FIXED_NOREPLACE for initial image mmap"
actually triggered EEXIST, which meant that probe_guest_base did not
do its job to select unused host virtual memory.  It's a mystery why
we have not seen larger problems because of this.

As I kept digging, I found quite a number of problems within
probe_guest_base and its subroutines.  I have rewritten it completely.
Hopefully it is much easier to understand in its new form.

Testing this has been difficult, because it is most visible with
non-PIE executables, and most modern distros default to PIE, and
our current implementation of --disable-pie does not work.


r~


Helge Deller (1):
  linux-user: Adjust initial brk when interpreter is close to executable

Richard Henderson (13):
  linux-user: Adjust task_unmapped_base for reserved_va
  linux-user: Define TASK_UNMAPPED_BASE in $guest/target_mman.h
  linux-user: Define ELF_ET_DYN_BASE in $guest/target_mman.h
  linux-user: Use MAP_FIXED_NOREPLACE for initial image mmap
  linux-user: Use elf_et_dyn_base for ET_DYN with interpreter
  linux-user: Do not adjust image mapping for host page size
  linux-user: Do not adjust zero_bss for host page size
  linux-user: Use zero_bss for PT_LOAD with no file contents too
  util/selfmap: Rewrite using qemu/interval-tree.h
  linux-user: Remove duplicate CPU_LOG_PAGE from probe_guest_base
  linux-user: Consolidate guest bounds check in probe_guest_base
  linux-user: Rewrite fixed probe_guest_base
  linux-user: Rewrite non-fixed probe_guest_base

 include/qemu/selfmap.h               |  20 +-
 linux-user/aarch64/target_mman.h     |  13 +
 linux-user/alpha/target_mman.h       |  11 +
 linux-user/arm/target_mman.h         |  11 +
 linux-user/cris/target_mman.h        |  12 +
 linux-user/hexagon/target_mman.h     |  13 +
 linux-user/hppa/target_mman.h        |   6 +
 linux-user/i386/target_mman.h        |  16 +
 linux-user/loongarch64/target_mman.h |  11 +
 linux-user/m68k/target_mman.h        |   5 +
 linux-user/microblaze/target_mman.h  |  11 +
 linux-user/mips/target_mman.h        |  10 +
 linux-user/nios2/target_mman.h       |  10 +
 linux-user/openrisc/target_mman.h    |  10 +
 linux-user/ppc/target_mman.h         |  20 +
 linux-user/qemu.h                    |   1 -
 linux-user/riscv/target_mman.h       |  10 +
 linux-user/s390x/target_mman.h       |  20 +
 linux-user/sh4/target_mman.h         |   7 +
 linux-user/sparc/target_mman.h       |  25 +
 linux-user/user-mmap.h               |   5 +-
 linux-user/x86_64/target_mman.h      |  15 +
 linux-user/xtensa/target_mman.h      |  10 +
 linux-user/elfload.c                 | 788 +++++++++++++--------------
 linux-user/main.c                    |  43 ++
 linux-user/mmap.c                    |  19 +-
 linux-user/syscall.c                 |  15 +-
 util/selfmap.c                       | 114 ++--
 28 files changed, 777 insertions(+), 474 deletions(-)

Comments

Alex Bennée Aug. 8, 2023, 5 p.m. UTC | #1
Richard Henderson <richard.henderson@linaro.org> writes:

> This is the second half of
>
> https://patchew.org/QEMU/20230804220032.295411-1-richard.henderson@linaro.org/
>
> which I held back because of regressions with s390x testing.
>
> It turns out that patch 4, "Use MAP_FIXED_NOREPLACE for initial image mmap"
> actually triggered EEXIST, which meant that probe_guest_base did not
> do its job to select unused host virtual memory.  It's a mystery why
> we have not seen larger problems because of this.
>
> As I kept digging, I found quite a number of problems within
> probe_guest_base and its subroutines.  I have rewritten it completely.
> Hopefully it is much easier to understand in its new form.
>
> Testing this has been difficult, because it is most visible with
> non-PIE executables, and most modern distros default to PIE, and
> our current implementation of --disable-pie does not work.

I should say:

  [alex@aarch64:~/l/q/b/ci.all.linux.static] review/linux-user-mapping-v4|… + retry.py -n 100 -c -- make check-tcg

Gives:

  Results summary:
  0: 100 times (100.00%), avg time 150.644 (14.47 varience/3.80 deviation)
  Ran command 100 times, 100 passes

which I think has squashed an intermittent bug plaguing CI so have a:

Tested-by: Alex Bennée <alex.bennee@linaro.org>

for the series.


>
>
> r~
>
>
> Helge Deller (1):
>   linux-user: Adjust initial brk when interpreter is close to executable
>
> Richard Henderson (13):
>   linux-user: Adjust task_unmapped_base for reserved_va
>   linux-user: Define TASK_UNMAPPED_BASE in $guest/target_mman.h
>   linux-user: Define ELF_ET_DYN_BASE in $guest/target_mman.h
>   linux-user: Use MAP_FIXED_NOREPLACE for initial image mmap
>   linux-user: Use elf_et_dyn_base for ET_DYN with interpreter
>   linux-user: Do not adjust image mapping for host page size
>   linux-user: Do not adjust zero_bss for host page size
>   linux-user: Use zero_bss for PT_LOAD with no file contents too
>   util/selfmap: Rewrite using qemu/interval-tree.h
>   linux-user: Remove duplicate CPU_LOG_PAGE from probe_guest_base
>   linux-user: Consolidate guest bounds check in probe_guest_base
>   linux-user: Rewrite fixed probe_guest_base
>   linux-user: Rewrite non-fixed probe_guest_base
>
>  include/qemu/selfmap.h               |  20 +-
>  linux-user/aarch64/target_mman.h     |  13 +
>  linux-user/alpha/target_mman.h       |  11 +
>  linux-user/arm/target_mman.h         |  11 +
>  linux-user/cris/target_mman.h        |  12 +
>  linux-user/hexagon/target_mman.h     |  13 +
>  linux-user/hppa/target_mman.h        |   6 +
>  linux-user/i386/target_mman.h        |  16 +
>  linux-user/loongarch64/target_mman.h |  11 +
>  linux-user/m68k/target_mman.h        |   5 +
>  linux-user/microblaze/target_mman.h  |  11 +
>  linux-user/mips/target_mman.h        |  10 +
>  linux-user/nios2/target_mman.h       |  10 +
>  linux-user/openrisc/target_mman.h    |  10 +
>  linux-user/ppc/target_mman.h         |  20 +
>  linux-user/qemu.h                    |   1 -
>  linux-user/riscv/target_mman.h       |  10 +
>  linux-user/s390x/target_mman.h       |  20 +
>  linux-user/sh4/target_mman.h         |   7 +
>  linux-user/sparc/target_mman.h       |  25 +
>  linux-user/user-mmap.h               |   5 +-
>  linux-user/x86_64/target_mman.h      |  15 +
>  linux-user/xtensa/target_mman.h      |  10 +
>  linux-user/elfload.c                 | 788 +++++++++++++--------------
>  linux-user/main.c                    |  43 ++
>  linux-user/mmap.c                    |  19 +-
>  linux-user/syscall.c                 |  15 +-
>  util/selfmap.c                       | 114 ++--
>  28 files changed, 777 insertions(+), 474 deletions(-)