[AMDGPU] Fix MC/Disassembler/AMDGPU/decode-err.txt. #96621
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
It fails downstream now that llvm#95237 removed flushing the output stream on printing every instruction.
|
@llvm/pr-subscribers-backend-amdgpu @llvm/pr-subscribers-mc Author: Ivan Kosarev (kosarev) ChangesIt fails downstream now that Full diff: https://github.com/llvm/llvm-project/pull/96621.diff 1 Files Affected:
diff --git a/llvm/test/MC/Disassembler/AMDGPU/decode-err.txt b/llvm/test/MC/Disassembler/AMDGPU/decode-err.txt
index d7417929a8e1b..3a2cef5feb318 100644
--- a/llvm/test/MC/Disassembler/AMDGPU/decode-err.txt
+++ b/llvm/test/MC/Disassembler/AMDGPU/decode-err.txt
@@ -1,17 +1,19 @@
-# RUN: llvm-mc -triple=amdgcn -mcpu=gfx900 -disassemble -show-encoding < %s 2>&1 | FileCheck -check-prefix=GCN %s
+# RUN: llvm-mc -triple=amdgcn -mcpu=gfx900 -disassemble -show-encoding < %s 2>&1 >/dev/null | FileCheck -check-prefix=GCN-ERR %s
# RUN: llvm-mc -triple=amdgcn -mcpu=gfx1100 -disassemble -show-encoding < %s 2>&1 | FileCheck -check-prefixes=GFX11,W32 %s
# RUN: llvm-mc -triple=amdgcn -mcpu=gfx1100 -mattr=-WavefrontSize32,+WavefrontSize64 -disassemble -show-encoding < %s 2>&1 | FileCheck -check-prefixes=GFX11,W64 %s
-# RUN: llvm-mc -triple=amdgcn -mcpu=gfx1200 -disassemble -show-encoding < %s 2>&1 | FileCheck -check-prefix=GFX12 %s
+# RUN: llvm-mc -triple=amdgcn -mcpu=gfx1100 -disassemble -show-encoding < %s 2>&1 >/dev/null | FileCheck -check-prefix=GFX11-ERR %s
+# RUN: llvm-mc -triple=amdgcn -mcpu=gfx1200 -disassemble -show-encoding < %s | FileCheck -check-prefix=GFX12 %s
+# RUN: llvm-mc -triple=amdgcn -mcpu=gfx1200 -disassemble -show-encoding < %s 2>&1 >/dev/null | FileCheck -check-prefix=GFX12-ERR %s
-# GCN: [[@LINE+1]]:1: warning: invalid instruction encoding
+# GCN-ERR: [[@LINE+1]]:1: warning: invalid instruction encoding
0xdf,0x00,0x00,0x02
# this is s_singleuse_vdst 0x1234, which is only valid on gfx1150
-# GFX11: [[@LINE+1]]:1: warning: invalid instruction encoding
+# GFX11-ERR: [[@LINE+1]]:1: warning: invalid instruction encoding
0x34,0x12,0x93,0xbf
# this is s_waitcnt_vscnt exec_hi, 0x1234, which is valid on gfx11, but not on gfx12
-# GFX12: [[@LINE+1]]:1: warning: invalid instruction encoding
+# GFX12-ERR: [[@LINE+1]]:1: warning: invalid instruction encoding
0x34,0x12,0x7f,0xbc
# W32: v_dual_add_f32 v5, 0xaf123456, v2 :: v_dual_fmaak_f32 v6, v3, v1, 0xaf123456 ; encoding: [0xff,0x04,0x02,0xc9,0x03,0x03,0x06,0x05,0x56,0x34,0x12,0xaf]
@@ -39,11 +41,11 @@
0x10,0x40,0x40,0xcc,0x00,0x11,0x02,0x18 # src2 sgpr0
# this is ds_add_f32 with gds bit which is not valid on gfx12+
-# GFX12: [[@LINE+1]]:1: warning: invalid instruction encoding
+# GFX12-ERR: [[@LINE+1]]:1: warning: invalid instruction encoding
0x00,0x00,0x56,0xd8,0x00,0x01,0x00,0x00
# this is image_msaa_load where samp field for gfx12 VSAMPLE is not all zeros
-# GFX12: [[@LINE+1]]:1: warning: invalid instruction encoding
+# GFX12-ERR: [[@LINE+1]]:1: warning: invalid instruction encoding
0x06,0x00,0x46,0xe4,0x01,0x10,0x80,0x00,0x05,0x06,0x07,0x00
# This is ds_read_b32 with gds bit which is not valid on gfx90a.
|
kosarev
commented
Jun 25, 2024
| 0x00,0x00,0x56,0xd8,0x00,0x01,0x00,0x00 | ||
|
|
||
| # this is image_msaa_load where samp field for gfx12 VSAMPLE is not all zeros | ||
| # GFX12: [[@LINE+1]]:1: warning: invalid instruction encoding | ||
| # GFX12-ERR: [[@LINE+1]]:1: warning: invalid instruction encoding | ||
| 0x06,0x00,0x46,0xe4,0x01,0x10,0x80,0x00,0x05,0x06,0x07,0x00 | ||
|
|
||
| # This is ds_read_b32 with gds bit which is not valid on gfx90a. |
There was a problem hiding this comment.
@jayfoad We don't seem to check GFX90A prefixes in this test.
There was a problem hiding this comment.
Yes, I noticed that recently thanks to filecheck_lint.py. If you add a RUN line for gfx90a, the check fails. It needs more investigation.
arsenm
reviewed
Jun 25, 2024
arsenm
approved these changes
Jun 25, 2024
arsenm
reviewed
Jun 25, 2024
| # RUN: llvm-mc -triple=amdgcn -mcpu=gfx1200 -disassemble -show-encoding < %s 2>&1 | FileCheck -check-prefix=GFX12 %s | ||
| # RUN: llvm-mc -triple=amdgcn -mcpu=gfx900 -disassemble -show-encoding -filetype=null < %s 2>&1 | FileCheck -check-prefix=GCN-ERR %s | ||
| # RUN: llvm-mc -triple=amdgcn -mcpu=gfx1100 -disassemble -show-encoding < %s 2>&1 | FileCheck -check-prefixes=W32 %s | ||
| # RUN: llvm-mc -triple=amdgcn -mcpu=gfx1100 -mattr=-WavefrontSize32,+WavefrontSize64 -disassemble -show-encoding < %s 2>&1 | FileCheck -check-prefixes=W64 %s |
There was a problem hiding this comment.
The capitalized wavefrontsize features shouldn't even parse? You also should never need to specify the negated form
There was a problem hiding this comment.
The subtarget features are processed as a state machine assigning values to the subtarget variable as each case is encountered. Code shouldn't be relying on a boolean is this feature present, but checking the resulting value of WavefrontSize
There was a problem hiding this comment.
Plenty of code goes straight to the feature bits:
$ grep -r getFeatureBits.*FeatureWavefrontSize lib/Target/AMDGPU/
lib/Target/AMDGPU/AsmParser/AMDGPUAsmParser.cpp: if (isGFX10Plus() && getFeatureBits()[AMDGPU::FeatureWavefrontSize64] &&
lib/Target/AMDGPU/AsmParser/AMDGPUAsmParser.cpp: !getFeatureBits()[AMDGPU::FeatureWavefrontSize32]) {
lib/Target/AMDGPU/AsmParser/AMDGPUAsmParser.cpp: if (!getFeatureBits()[AMDGPU::FeatureWavefrontSize32])
lib/Target/AMDGPU/AsmParser/AMDGPUAsmParser.cpp: if (!getFeatureBits()[AMDGPU::FeatureWavefrontSize64])
lib/Target/AMDGPU/AsmParser/AMDGPUAsmParser.cpp: if (!getFeatureBits()[AMDGPU::FeatureWavefrontSize32])
lib/Target/AMDGPU/AsmParser/AMDGPUAsmParser.cpp: if (!getFeatureBits()[AMDGPU::FeatureWavefrontSize64])
lib/Target/AMDGPU/MCTargetDesc/AMDGPUMCKernelDescriptor.cpp: if (STI->getFeatureBits().test(FeatureWavefrontSize32))
lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp: if (STI->getFeatureBits().test(FeatureWavefrontSize16))
lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp: if (STI->getFeatureBits().test(FeatureWavefrontSize32))
lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp: STI->getFeatureBits().test(FeatureWavefrontSize32);
lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp: STI->getFeatureBits().test(FeatureWavefrontSize32);
lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp: bool IsWave32 = STI->getFeatureBits().test(FeatureWavefrontSize32);
lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp: if (STI->getFeatureBits().test(FeatureWavefrontSize32)) {
There was a problem hiding this comment.
The capitalized wavefrontsize features shouldn't even parse?
It seems to get lowered before processing. Not sure if it should or shouldn't be done, though.
Re negations: that would be an unrelated change anyway, so I will submit this as is, if no objections.
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/123/builds/747 Here is the relevant piece of the build log for the reference: |
lravenclaw
pushed a commit
to lravenclaw/llvm-project
that referenced
this pull request
Jul 3, 2024
It fails downstream now that llvm#95237 removed flushing the output stream on printing every instruction.
AlexisPerry
pushed a commit
to llvm-project-tlp/llvm-project
that referenced
this pull request
Jul 9, 2024
It fails downstream now that llvm#95237 removed flushing the output stream on printing every instruction.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
It fails downstream now that
#95237 removed flushing the output stream on printing every instruction.