Looks good to me.
/Erik
On 2018-11-28 13:17, Jakub Vaněk wrote:
Hi David,
On 2018-11-26 at 17:36 +1000, David Holmes wrote:
Hi Jakub,
On 26/11/2018 1:53 am, Jakub Vaněk wrote:
Hi,
This patch raises the minimum architectural level for ARM CPUs to
ARMv5TE. It is done through changing the -march flag in the current
CPU-specific CFLAGS and ASFLAGS.
This patch depends on "Append assembler flags on ARM targets" patch
for
ASFLAGS handling.
Reason for this change: assembler code in linux_arm_32.s uses PLD
instructions. ARM ISA manual mentions this: "This instruction is
available in E variants of ARM architecture v5 and above." (thanks
to
David Holmes for noticing this).
FYI I did a bit of digging and we tried to make this change back in
2010
and ran into problems:
"It appears that gcc assumes that if you choose the armv5te
architecture
option, it can use the "E" instruction extensions which provide 64
bit
load and stores to 4 byte aligned addresses. I tried two different
compiler versions 4.1.2 and 4.2.3 and they both had the same
problem.
The <hardware> does support ldrd/strd but they do not support 4 byte
aligned operations. The addresses must be 8 byte aligned."
How things stand today I can not say.
I have searched for this and indeed, this bug wasn't fixed until
recently: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82445
For our own use, it seems that Ubuntu gcc7 is patched, whether Debian
gcc6 isn't. I think that the best solution is to keep armv5t in CFLAGS
and only raise armv5te in ASFLAGS, where it is actually needed. This
way GCC still generates older non-buggy code and the assembler works
with pld. I have tested this configuration and it compiles
successfully.
https://ci.adoptopenjdk.net/view/ev3dev/job/openjdk12_build_ev3_linux/40/
Thanks,
Jakub
Cheers,
David
# HG changeset patch
# User Jakub Vaněk <linuxtar...@gmail.com>
# Date 1543252130 -3600
# Mon Nov 26 18:08:50 2018 +0100
# Node ID bfd9e7032cfdc8c814ddaf4bb471a014fbbf1a89
# Parent d2db3b32d5614029f2b1af2731c0d9ec10d88767
Fix undefined pld instruction on arm-sflt
diff --git a/make/autoconf/flags.m4 b/make/autoconf/flags.m4
--- a/make/autoconf/flags.m4
+++ b/make/autoconf/flags.m4
@@ -46,6 +46,11 @@
AC_MSG_CHECKING([for ABI profle])
AC_MSG_RESULT([$OPENJDK_TARGET_ABI_PROFILE])
+ # --- Arm-sflt CFLAGS and ASFLAGS ---
+ # Armv5te is required for assembler, because pld insn used in arm32
hotspot is only in v5E and above.
+ # However, there is also a GCC bug which generates unaligned strd/ldrd
instructions on armv5te:
+ # https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82445, and it was fixed
only quite recently.
+ # The resulting compromise is to enable v5TE for assembler and let GCC
generate code for v5T.
if test "x$OPENJDK_TARGET_ABI_PROFILE" = xarm-vfp-sflt; then
ARM_FLOAT_TYPE=vfp-sflt
ARM_ARCH_TYPE_FLAGS='-march=armv7-a -mthumb'
@@ -57,11 +62,11 @@
elif test "x$OPENJDK_TARGET_ABI_PROFILE" = xarm-sflt; then
ARM_FLOAT_TYPE=sflt
ARM_ARCH_TYPE_FLAGS='-march=armv5t -marm'
- ARM_ARCH_TYPE_ASFLAGS='-march=armv5t'
+ ARM_ARCH_TYPE_ASFLAGS='-march=armv5te'
elif test "x$OPENJDK_TARGET_ABI_PROFILE" = xarmv5-vfp-sflt; then
ARM_FLOAT_TYPE=vfp-sflt
ARM_ARCH_TYPE_FLAGS='-march=armv5t -marm'
- ARM_ARCH_TYPE_ASFLAGS='-march=armv5t'
+ ARM_ARCH_TYPE_ASFLAGS='-march=armv5te'
elif test "x$OPENJDK_TARGET_ABI_PROFILE" = xarmv6-vfp-hflt; then
ARM_FLOAT_TYPE=vfp-hflt
ARM_ARCH_TYPE_FLAGS='-march=armv6 -marm'
