Skip to content

[SCEV] Make sure starting block is marked as visited when recursively collecting loop guards. #120749

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

Merged
merged 1 commit into from
Dec 31, 2024

Conversation

juliannagele
Copy link
Member

When collectFromBlock is called without a predecessor (in particular for loops that don't have a unique predecessor outside the loop) we never start climbing the predecessor chain, and thus don't mark the starting block as visited.

Fixes #120615.

@juliannagele juliannagele requested a review from nikic as a code owner December 20, 2024 15:36
@juliannagele juliannagele requested a review from fhahn December 20, 2024 15:36
@llvmbot llvmbot added the llvm:analysis Includes value tracking, cost tables and constant folding label Dec 20, 2024
@llvmbot
Copy link
Member

llvmbot commented Dec 20, 2024

@llvm/pr-subscribers-llvm-analysis

Author: Julian Nagele (juliannagele)

Changes

When collectFromBlock is called without a predecessor (in particular for loops that don't have a unique predecessor outside the loop) we never start climbing the predecessor chain, and thus don't mark the starting block as visited.

Fixes #120615.


Full diff: https://github.com/llvm/llvm-project/pull/120749.diff

2 Files Affected:

  • (modified) llvm/lib/Analysis/ScalarEvolution.cpp (+1)
  • (modified) llvm/test/Analysis/ScalarEvolution/backedge-taken-count-guard-info-with-multiple-predecessors.ll (+26)
diff --git a/llvm/lib/Analysis/ScalarEvolution.cpp b/llvm/lib/Analysis/ScalarEvolution.cpp
index 1e4bb1d606cd39..91bbe92a94c210 100644
--- a/llvm/lib/Analysis/ScalarEvolution.cpp
+++ b/llvm/lib/Analysis/ScalarEvolution.cpp
@@ -15754,6 +15754,7 @@ void ScalarEvolution::LoopGuards::collectFromBlock(
   // original header.
   // TODO: share this logic with isLoopEntryGuardedByCond.
   unsigned NumCollectedConditions = 0;
+  VisitedBlocks.insert(Block);
   std::pair<const BasicBlock *, const BasicBlock *> Pair(Pred, Block);
   for (; Pair.first;
        Pair = SE.getPredecessorWithUniqueSuccessorForBB(Pair.first)) {
diff --git a/llvm/test/Analysis/ScalarEvolution/backedge-taken-count-guard-info-with-multiple-predecessors.ll b/llvm/test/Analysis/ScalarEvolution/backedge-taken-count-guard-info-with-multiple-predecessors.ll
index 81fe96a2f30c09..46dccf454f21ac 100644
--- a/llvm/test/Analysis/ScalarEvolution/backedge-taken-count-guard-info-with-multiple-predecessors.ll
+++ b/llvm/test/Analysis/ScalarEvolution/backedge-taken-count-guard-info-with-multiple-predecessors.ll
@@ -310,3 +310,29 @@ inner.header:
 exit:
   ret void
 }
+
+; Checks correct traversal for loops without a unique predecessor
+; outside the loop.
+define void @pr120615() {
+; CHECK-LABEL: pr120615
+; CHECK-NEXT:  Determining loop execution counts for: @pr120615
+; CHECK-NEXT:  Loop %header: backedge-taken count is i32 0
+; CHECK-NEXT:  Loop %header: constant max backedge-taken count is i32 0
+; CHECK-NEXT:  Loop %header: symbolic max backedge-taken count is i32 0
+; CHECK-NEXT:  Loop %header: Trip multiple is 1
+entry:
+  br label %header
+
+bb:
+  br label %header
+
+header:
+  %0 = phi i32 [ %1, %header ], [ 0, %bb ], [ 0, %entry ]
+  %1 = add i32 %0, 1
+  %icmp = icmp slt i32 %0, 0
+  br i1 %icmp, label %header, label %exit
+
+exit:
+  ret void
+
+}

Copy link
Contributor

@fhahn fhahn left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM thanks

@danilaml
Copy link
Collaborator

LGTM as well. Can we merge this?

@juliannagele juliannagele merged commit f035351 into llvm:main Dec 31, 2024
10 checks passed
@llvm-ci
Copy link
Collaborator

llvm-ci commented Dec 31, 2024

LLVM Buildbot has detected a new failure on builder openmp-offload-libc-amdgpu-runtime running on omp-vega20-1 while building llvm at step 7 "Add check check-offload".

Full details are available at: https://lab.llvm.org/buildbot/#/builders/73/builds/10995

Here is the relevant piece of the build log for the reference
Step 7 (Add check check-offload) failure: test (failure)
******************** TEST 'libomptarget :: amdgcn-amd-amdhsa :: sanitizer/ptr_outside_alloc_2.c' FAILED ********************
Exit Code: 1

Command Output (stdout):
--
# RUN: at line 2
/home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/./bin/clang -fopenmp    -I /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.src/offload/test -I /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/runtimes/runtimes-bins/openmp/runtime/src -L /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/runtimes/runtimes-bins/offload -L /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/./lib -L /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/runtimes/runtimes-bins/openmp/runtime/src  -nogpulib -Wl,-rpath,/home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/runtimes/runtimes-bins/offload -Wl,-rpath,/home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/runtimes/runtimes-bins/openmp/runtime/src -Wl,-rpath,/home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/./lib  -fopenmp-targets=amdgcn-amd-amdhsa -O3 /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.src/offload/test/sanitizer/ptr_outside_alloc_2.c -o /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/runtimes/runtimes-bins/offload/test/amdgcn-amd-amdhsa/sanitizer/Output/ptr_outside_alloc_2.c.tmp -Xoffload-linker -lc -Xoffload-linker -lm /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/./lib/libomptarget.devicertl.a
# executed command: /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/./bin/clang -fopenmp -I /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.src/offload/test -I /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/runtimes/runtimes-bins/openmp/runtime/src -L /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/runtimes/runtimes-bins/offload -L /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/./lib -L /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/runtimes/runtimes-bins/openmp/runtime/src -nogpulib -Wl,-rpath,/home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/runtimes/runtimes-bins/offload -Wl,-rpath,/home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/runtimes/runtimes-bins/openmp/runtime/src -Wl,-rpath,/home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/./lib -fopenmp-targets=amdgcn-amd-amdhsa -O3 /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.src/offload/test/sanitizer/ptr_outside_alloc_2.c -o /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/runtimes/runtimes-bins/offload/test/amdgcn-amd-amdhsa/sanitizer/Output/ptr_outside_alloc_2.c.tmp -Xoffload-linker -lc -Xoffload-linker -lm /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/./lib/libomptarget.devicertl.a
# RUN: at line 3
/home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/./bin/not --crash env -u LLVM_DISABLE_SYMBOLIZATION OFFLOAD_TRACK_ALLOCATION_TRACES=1 /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/runtimes/runtimes-bins/offload/test/amdgcn-amd-amdhsa/sanitizer/Output/ptr_outside_alloc_2.c.tmp 2>&1 | /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/./bin/FileCheck /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.src/offload/test/sanitizer/ptr_outside_alloc_2.c --check-prefixes=CHECK
# executed command: /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/./bin/not --crash env -u LLVM_DISABLE_SYMBOLIZATION OFFLOAD_TRACK_ALLOCATION_TRACES=1 /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/runtimes/runtimes-bins/offload/test/amdgcn-amd-amdhsa/sanitizer/Output/ptr_outside_alloc_2.c.tmp
# executed command: /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.build/./bin/FileCheck /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.src/offload/test/sanitizer/ptr_outside_alloc_2.c --check-prefixes=CHECK
# .---command stderr------------
# | /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.src/offload/test/sanitizer/ptr_outside_alloc_2.c:21:11: error: CHECK: expected string not found in input
# | // CHECK: OFFLOAD ERROR: Memory access fault by GPU {{.*}} (agent 0x{{.*}}) at virtual address [[PTR:0x[0-9a-z]*]]. Reasons: {{.*}}
# |           ^
# | <stdin>:1:1: note: scanning from here
# | AMDGPU error: Error in hsa_amd_memory_pool_allocate: HSA_STATUS_ERROR_OUT_OF_RESOURCES: The runtime failed to allocate the necessary resources. This error may also occur when the core runtime library needs to spawn threads or create internal OS-specific events.
# | ^
# | 
# | Input file: <stdin>
# | Check file: /home/ompworker/bbot/openmp-offload-libc-amdgpu-runtime/llvm.src/offload/test/sanitizer/ptr_outside_alloc_2.c
# | 
# | -dump-input=help explains the following input dump.
# | 
# | Input was:
# | <<<<<<
# |           1: AMDGPU error: Error in hsa_amd_memory_pool_allocate: HSA_STATUS_ERROR_OUT_OF_RESOURCES: The runtime failed to allocate the necessary resources. This error may also occur when the core runtime library needs to spawn threads or create internal OS-specific events. 
# | check:21     X~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ error: no match found
# |           2: AMDGPU error: Error in hsa_amd_memory_pool_allocate: HSA_STATUS_ERROR_OUT_OF_RESOURCES: The runtime failed to allocate the necessary resources. This error may also occur when the core runtime library needs to spawn threads or create internal OS-specific events. 
# | check:21     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# |           3: "PluginInterface" error: Failure to allocate device memory: Failed to allocate from memory manager 
# | check:21     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# |           4: omptarget error: Call to getTargetPointer returned null pointer (device failure or illegal mapping). 
# | check:21     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# |           5: omptarget error: Call to targetDataBegin failed, abort target. 
# | check:21     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# |           6: omptarget error: Failed to process data before launching the kernel. 
# | check:21     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# |           .
# |           .
# |           .
# | >>>>>>
# `-----------------------------
# error: command failed with exit status: 1

--

********************


juliannagele added a commit to juliannagele/llvm-project that referenced this pull request Jan 6, 2025
… collecting loop guards. (llvm#120749)

When `collectFromBlock` is called without a predecessor (in particular
for loops that don't have a unique predecessor outside the loop) we
never start climbing the predecessor chain, and thus don't mark the
starting block as visited.

Fixes llvm#120615.

(cherry picked from commit f035351)
juliannagele added a commit to swiftlang/llvm-project that referenced this pull request Jan 6, 2025
…9797)

* [SCEV] Fix exit condition for recursive loop guard collection (llvm#120442)

When assumptions are present `Terms.size()` does not actually count the
number of conditions collected from dominating branches; introduce a
separate counter.

Fixes llvm#120237

(cherry picked from commit acfd26a)

* [SCEV] Make sure starting block is marked as visited when recursively collecting loop guards. (llvm#120749)

When `collectFromBlock` is called without a predecessor (in particular
for loops that don't have a unique predecessor outside the loop) we
never start climbing the predecessor chain, and thus don't mark the
starting block as visited.

Fixes llvm#120615.

(cherry picked from commit f035351)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
llvm:analysis Includes value tracking, cost tables and constant folding
Projects
None yet
Development

Successfully merging this pull request may close these issues.

[SCEV] Segfault in SCEV LoopGuards
5 participants