Alternate fix for assertion on armv7hl (RHBZ#1999878)

This commit is contained in:
Richard W.M. Jones 2021-09-12 20:45:17 +01:00
parent c5273992ba
commit 65ccb9e09a
3 changed files with 78 additions and 34 deletions

View File

@ -0,0 +1,73 @@
From 1331e4eec016a295949009b4360c592401b089f7 Mon Sep 17 00:00:00 2001
From: Richard Henderson <richard.henderson@linaro.org>
Date: Sun, 12 Sep 2021 10:49:25 -0700
Subject: [PATCH] tcg/arm: Reduce vector alignment requirement for NEON
With arm32, the ABI gives us 8-byte alignment for the stack.
While it's possible to realign the stack to provide 16-byte alignment,
it's far easier to simply not encode 16-byte alignment in the
VLD1 and VST1 instructions that we emit.
Remove the assertion in temp_allocate_frame, limit natural alignment
to the provided stack alignment, and add a comment.
Reported-by: Richard W.M. Jones <rjones@redhat.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
tcg/arm/tcg-target.c.inc | 13 +++++++++----
tcg/tcg.c | 8 +++++++-
2 files changed, 16 insertions(+), 5 deletions(-)
diff --git a/tcg/arm/tcg-target.c.inc b/tcg/arm/tcg-target.c.inc
index 007ceee68e..34acfb522c 100644
--- a/tcg/arm/tcg-target.c.inc
+++ b/tcg/arm/tcg-target.c.inc
@@ -2477,8 +2477,13 @@ static void tcg_out_ld(TCGContext *s, TCGType type, TCGReg arg,
tcg_out_vldst(s, INSN_VLD1 | 0x7d0, arg, arg1, arg2);
return;
case TCG_TYPE_V128:
- /* regs 2; size 8; align 16 */
- tcg_out_vldst(s, INSN_VLD1 | 0xae0, arg, arg1, arg2);
+ /*
+ * We have only 8-byte alignment for the stack per the ABI.
+ * Rather than dynamically re-align the stack, it's easier
+ * to simply not request alignment beyond that. So:
+ * regs 2; size 8; align 8
+ */
+ tcg_out_vldst(s, INSN_VLD1 | 0xad0, arg, arg1, arg2);
return;
default:
g_assert_not_reached();
@@ -2497,8 +2502,8 @@ static void tcg_out_st(TCGContext *s, TCGType type, TCGReg arg,
tcg_out_vldst(s, INSN_VST1 | 0x7d0, arg, arg1, arg2);
return;
case TCG_TYPE_V128:
- /* regs 2; size 8; align 16 */
- tcg_out_vldst(s, INSN_VST1 | 0xae0, arg, arg1, arg2);
+ /* See tcg_out_ld re alignment: regs 2; size 8; align 8 */
+ tcg_out_vldst(s, INSN_VST1 | 0xad0, arg, arg1, arg2);
return;
default:
g_assert_not_reached();
diff --git a/tcg/tcg.c b/tcg/tcg.c
index 4142d42d77..ca5bcc4635 100644
--- a/tcg/tcg.c
+++ b/tcg/tcg.c
@@ -3060,7 +3060,13 @@ static void temp_allocate_frame(TCGContext *s, TCGTemp *ts)
g_assert_not_reached();
}
- assert(align <= TCG_TARGET_STACK_ALIGN);
+ /*
+ * Assume the stack is sufficiently aligned.
+ * This affects e.g. ARM NEON, where we have 8 byte stack alignment
+ * and do not require 16 byte vector alignment. This seems slightly
+ * easier than fully parameterizing the above switch statement.
+ */
+ align = MIN(TCG_TARGET_STACK_ALIGN, align);
off = ROUND_UP(s->current_frame_offset, align);
/* If we've exhausted the stack frame, restart with a smaller TB. */
--
2.32.0

View File

@ -1,32 +0,0 @@
From 319935db129085bf4a3886dfd5662c687e6ce814 Mon Sep 17 00:00:00 2001
From: "Richard W.M. Jones" <rjones@redhat.com>
Date: Wed, 1 Sep 2021 17:12:20 +0100
Subject: [PATCH 2/2] tcg/arm: Increase stack alignment for function generation
This avoids the following assertion when the kernel initializes X.509
certificates:
[ 7.315373] Loading compiled-in X.509 certificates
qemu-system-arm: ../tcg/tcg.c:3063: temp_allocate_frame: Assertion `align <= TCG_TARGET_STACK_ALIGN' failed.
Signed-off-by: Richard W.M. Jones <rjones@redhat.com>
---
tcg/arm/tcg-target.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/tcg/arm/tcg-target.h b/tcg/arm/tcg-target.h
index d113b7f8db..5c8f3f8c55 100644
--- a/tcg/arm/tcg-target.h
+++ b/tcg/arm/tcg-target.h
@@ -115,7 +115,7 @@ extern bool use_neon_instructions;
#endif
/* used for function call generation */
-#define TCG_TARGET_STACK_ALIGN 8
+#define TCG_TARGET_STACK_ALIGN 16
#define TCG_TARGET_CALL_ALIGN_ARGS 1
#define TCG_TARGET_CALL_STACK_OFFSET 0
--
2.32.0

View File

@ -282,7 +282,7 @@ Obsoletes: %{name}-system-unicore32-core <= %{epoch}:%{version}-%{release}
Summary: QEMU is a FAST! processor emulator Summary: QEMU is a FAST! processor emulator
Name: qemu Name: qemu
Version: 6.1.0 Version: 6.1.0
Release: 6%{?rcrel}%{?dist} Release: 7%{?rcrel}%{?dist}
Epoch: 2 Epoch: 2
License: GPLv2 and BSD and MIT and CC-BY License: GPLv2 and BSD and MIT and CC-BY
URL: http://www.qemu.org/ URL: http://www.qemu.org/
@ -306,7 +306,7 @@ Patch1: 0001-target-i386-add-missing-bits-to-CR4_RESERVED_MASK.patch
# Fix assertion on armv7hl # Fix assertion on armv7hl
# https://bugzilla.redhat.com/show_bug.cgi?id=1999878 # https://bugzilla.redhat.com/show_bug.cgi?id=1999878
Patch2: 0002-tcg-arm-Increase-stack-alignment-for-function-genera.patch Patch2: 0001-tcg-arm-Reduce-vector-alignment-requirement-for-NEON.patch
BuildRequires: meson >= %{meson_version} BuildRequires: meson >= %{meson_version}
BuildRequires: zlib-devel BuildRequires: zlib-devel
@ -2236,6 +2236,9 @@ useradd -r -u 107 -g qemu -G kvm -d / -s /sbin/nologin \
%changelog %changelog
* Sun Sep 12 2021 Richard W.M. Jones <rjones@redhat.com> - 6.1.0-7
- Alternate fix for assertion on armv7hl (RHBZ#1999878)
* Wed Sep 01 2021 Richard W.M. Jones <rjones@redhat.com> - 6.1.0-6 * Wed Sep 01 2021 Richard W.M. Jones <rjones@redhat.com> - 6.1.0-6
- Fix assertion on armv7hl (RHBZ#1999878) - Fix assertion on armv7hl (RHBZ#1999878)