]> git.ipfire.org Git - thirdparty/qemu.git/commitdiff
linux-user/microblaze: Fix little-endianness binary
authorPhilippe Mathieu-Daudé <philmd@linaro.org>
Mon, 6 Oct 2025 15:36:31 +0000 (17:36 +0200)
committerMichael Tokarev <mjt@tls.msk.ru>
Fri, 17 Oct 2025 06:04:12 +0000 (09:04 +0300)
MicroBlaze CPU model has a "little-endian" property, pointing to
the @endi internal field. Commit c36ec3a9655 ("hw/microblaze:
Explicit CPU endianness") took care of having all MicroBlaze
boards with an explicit default endianness, so later commit
415aae543ed ("target/microblaze: Consider endianness while
translating code") could infer the endianness at runtime from
the @endi field, and not a compile time via the TARGET_BIG_ENDIAN
definition. Doing so, we forgot to make the endianness explicit
on user emulation, so there all CPUs are started with the default
"little-endian=off" value, leading to breaking support for little
endian binaries:

  $ readelf -h ./hello-world-mbel
  ELF Header:
    Magic:   7f 45 4c 46 01 01 01 00 00 00 00 00 00 00 00 00
    Class:                             ELF32
    Data:                              2's complement, little endian

  $ qemu-microblazeel ./hello-world-mbel
  qemu: uncaught target signal 11 (Segmentation fault) - core dumped
  Segmentation fault (core dumped)

Fix by restoring the previous behavior of starting with the
builtin endianness of the binary:

  $ qemu-microblazeel ./hello-world-mbel
  Hello World

Cc: qemu-stable@nongnu.org
Fixes: 415aae543ed ("target/microblaze: Consider endianness while translating code")
Reported-by: Edgar E. Iglesias <edgar.iglesias@gmail.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
Message-Id: <20251006173350.17455-1-philmd@linaro.org>
(cherry picked from commit 91fc6d8101de97c588e0a4263cf4f6148b3e702a)
(Mjt: adapt for missing v10.1.0-38-gaf880af8d4
 "linux-user: Move get_elf_cpu_model to target/elfload.c")
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
linux-user/microblaze/target_elf.h

index 8a8f1debff9e7cf78af9ba22bb0b9adc032e4aff..cab4c9187cc128470fe127b4b488489523ac54cd 100644 (file)
@@ -9,6 +9,7 @@
 #define MICROBLAZE_TARGET_ELF_H
 static inline const char *cpu_get_model(uint32_t eflags)
 {
-    return "any";
+    return TARGET_BIG_ENDIAN ? "any,little-endian=off"
+                             : "any,little-endian=on";
 }
 #endif