-
Notifications
You must be signed in to change notification settings - Fork 12.3k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[llvm][ARM] Correct the properties of trap instructions #113287
Conversation
Fixes llvm#113154 The encodings used for llvm.trap() on ARM were all marked as barriers and terminators. This lead to stack frame destroy code being inserted before the trap if the trap was the last thing in the function and it had no return statement. ``` void fn() { volatile int i = 0; __builtin_trap(); } ``` Produced: ``` fn: push {r11, lr} << stack frame create <...> mov sp, r11 pop {r11, lr} << stack frame destroy .inst 0xe7ffdefe << trap bx lr ``` All the other targets don't mark them this way, instead they mark them with isTrap. I've changed ARM to do this, which fixes the code generation: ``` fn: push {r11, lr} << stack frame create <...> .inst 0xe7ffdefe << trap mov sp, r11 pop {r11, lr} << stack frame destroy bx lr ``` I've updated the existing trap test to force the need for a stack frame, then check that the instruction immediately after the trap is resetting the stack pointer.
@llvm/pr-subscribers-backend-arm Author: David Spickett (DavidSpickett) ChangesFixes #113154 The encodings used for llvm.trap() on ARM were all marked as barriers and terminators. This lead to stack frame destroy code being inserted before the trap if the trap was the last thing in the function and it had no return statement.
Produced:
All the other targets don't mark them this way, instead they mark them with isTrap. I've changed ARM to do this, which fixes the code generation:
I've updated the existing trap test to force the need for a stack frame, then check that the instruction immediately after the trap is resetting the stack pointer. Full diff: https://github.com/llvm/llvm-project/pull/113287.diff 3 Files Affected:
diff --git a/llvm/lib/Target/ARM/ARMInstrInfo.td b/llvm/lib/Target/ARM/ARMInstrInfo.td
index ed68c6ff20cde5..d24d4af36f0d86 100644
--- a/llvm/lib/Target/ARM/ARMInstrInfo.td
+++ b/llvm/lib/Target/ARM/ARMInstrInfo.td
@@ -2378,13 +2378,13 @@ def UDF : AInoP<(outs), (ins imm0_65535:$imm16), MiscFrm, NoItinerary,
* - In ARM: UDF #60896;
* - In Thumb: UDF #254 followed by a branch-to-self.
*/
-let isBarrier = 1, isTerminator = 1 in
+let isTrap = 1 in
def TRAPNaCl : AXI<(outs), (ins), MiscFrm, NoItinerary,
"trap", [(trap)]>,
Requires<[IsARM,UseNaClTrap]> {
let Inst = 0xe7fedef0;
}
-let isBarrier = 1, isTerminator = 1 in
+let isTrap = 1 in
def TRAP : AXI<(outs), (ins), MiscFrm, NoItinerary,
"trap", [(trap)]>,
Requires<[IsARM,DontUseNaClTrap]> {
diff --git a/llvm/lib/Target/ARM/ARMInstrThumb.td b/llvm/lib/Target/ARM/ARMInstrThumb.td
index 2ad78f8cd8d4c7..b92f42874bbddb 100644
--- a/llvm/lib/Target/ARM/ARMInstrThumb.td
+++ b/llvm/lib/Target/ARM/ARMInstrThumb.td
@@ -675,7 +675,7 @@ def tSVC : T1pI<(outs), (ins imm0_255:$imm), IIC_Br,
}
// The assembler uses 0xDEFE for a trap instruction.
-let isBarrier = 1, isTerminator = 1 in
+let isTrap = 1 in
def tTRAP : TI<(outs), (ins), IIC_Br,
"trap", [(trap)]>, Encoding16, Sched<[WriteBr]> {
let Inst = 0xdefe;
diff --git a/llvm/test/CodeGen/ARM/trap.ll b/llvm/test/CodeGen/ARM/trap.ll
index fe046e70166553..a5d2b2081f28f3 100644
--- a/llvm/test/CodeGen/ARM/trap.ll
+++ b/llvm/test/CodeGen/ARM/trap.ll
@@ -1,3 +1,6 @@
+;; Check encodings of trap instructions and that their properties are set
+;; correctly so that they are not placed after the stack frame is destroyed.
+
; RUN: llc < %s -mtriple=arm-apple-darwin | FileCheck %s -check-prefix=DARWIN
; RUN: llc < %s -mtriple=arm-apple-darwin -trap-func=_trap | FileCheck %s -check-prefix=FUNC
; RUN: llc < %s -mtriple=arm-apple-darwin -trap-func=_trap -O0 | FileCheck %s -check-prefix=FUNC
@@ -29,22 +32,31 @@
; rdar://7961298
; rdar://9249183
-define void @t() nounwind {
+define void @t() noinline optnone {
entry:
+ ;; So that we have a stack frame.
+ %1 = alloca i32, align 4
+ store volatile i32 0, ptr %1, align 4
+
; DARWIN-LABEL: t:
-; DARWIN: trap
+; DARWIN: trap
+; DARWIN-NEXT: add sp, sp, #4
; FUNC-LABEL: t:
-; FUNC: bl __trap
+; FUNC: bl __trap
+; FUNC-NEXT: add sp, sp, #4
; NACL-LABEL: t:
-; NACL: .inst 0xe7fedef0
+; NACL: .inst 0xe7fedef0
+; NACL-NEXT: add sp, sp, #4
; ARM-LABEL: t:
-; ARM: .inst 0xe7ffdefe
+; ARM: .inst 0xe7ffdefe
+; ARM-NEXT: add sp, sp, #4
; THUMB-LABEL: t:
-; THUMB: .inst.n 0xdefe
+; THUMB: .inst.n 0xdefe
+; THUMB-NEXT: add sp, #4
; ENCODING-NACL: e7fedef0 trap
@@ -53,25 +65,34 @@ entry:
; ENCODING-THUMB: defe trap
call void @llvm.trap()
- unreachable
+ ret void
}
-define void @t2() nounwind {
+define void @t2() {
entry:
+ ;; So that we have a stack frame.
+ %1 = alloca i32, align 4
+ store volatile i32 0, ptr %1, align 4
+
; DARWIN-LABEL: t2:
-; DARWIN: udf #254
+; DARWIN: udf #254
+; DARWIN-NEXT: add sp, sp, #4
; FUNC-LABEL: t2:
-; FUNC: bl __trap
+; FUNC: bl __trap
+; FUNC-NEXT: add sp, sp, #4
; NACL-LABEL: t2:
-; NACL: bkpt #0
+; NACL: bkpt #0
+; NACL-NEXT: add sp, sp, #4
; ARM-LABEL: t2:
-; ARM: bkpt #0
+; ARM: bkpt #0
+; ARM-NEXT: add sp, sp, #4
; THUMB-LABEL: t2:
-; THUMB: bkpt #0
+; THUMB: bkpt #0
+; THUMB-NEXT: add sp, #4
; ENCODING-NACL: e1200070 bkpt #0
@@ -80,7 +101,7 @@ entry:
; ENCODING-THUMB: be00 bkpt #0
call void @llvm.debugtrap()
- unreachable
+ ret void
}
declare void @llvm.trap() nounwind
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
(It looks like this was discovered before in https://reviews.llvm.org/D53614, but it wasn't fixed at the time for some reason.)
Fixes #113154
The encodings used for llvm.trap() on ARM were all marked as barriers and terminators. This lead to stack frame destroy code being inserted before the trap if the trap was the last thing in the function and it had no return statement.
Produced:
All the other targets don't mark them this way, instead they mark them with isTrap. I've changed ARM to do this, which fixes the code generation:
I've updated the existing trap test to force the need for a stack frame, then check that the instruction immediately after the trap is resetting the stack pointer.
debugtrap was already working but I've added the same checks for it anyway.