Regression/commandlist/gpu.cpp flaky fail on AMD on unrelated changes

[ayylol]

Describe the bug

https://github.com/intel/llvm/actions/runs/13596042579/job/38014872400?pr=17200

-- Testing: 2268 tests, 10 workers --
FAIL: SYCL :: Regression/commandlist/gpu.cpp (2265 of 2268)
******************** TEST 'SYCL :: Regression/commandlist/gpu.cpp' FAILED ********************
Exit Code: 134

Command Output (stdout):
--
NThreads: GPU 8
   shared Queue: true
   array size: 1048576 (1<<20)
   SyCL items: 1048576
   iter GPU:   200
selected dev: AMD Radeon PRO W6800


                           subm   strt    end    run
GPU[3]:  nPrimes: 82024    0.00   0.05   3.58   3.53      0.00   3.58   3.58
GPU[1]:  nPrimes: 82024    3.58   3.58   6.69   3.11      0.00   6.69   6.69
GPU[0]:  nPrimes: 82024    6.69   6.72  12.67   5.95      0.00  12.67  12.67
GPU[4]:  nPrimes: 82024   12.68  12.72  19.06   6.34      0.00  19.53  19.53
GPU[2]:  nPrimes: 82024   19.30  19.53  28.75   9.22      0.00  28.76  28.76
GPU[6]:  nPrimes: 82024   28.76  28.76  30.80   2.04      0.00  30.82  30.82
GPU[7]:  nPrimes: 82024   30.82  30.82  31.75   0.92      0.00  31.75  31.75
GPU[5]:  nPrimes: 82024   31.75  31.75  32.01   0.26      0.00  32.01  32.01
GPU average time: 3.92
total time: 32.02
Passed

--
Command Output (stderr):
--
RUN: at line 4: env UR_HIP_ENABLE_IMAGE_SUPPORT=1 ONEAPI_DEVICE_SELECTOR=hip:gpu  /__w/llvm/llvm/build-e2e/Regression/commandlist/Output/gpu.cpp.tmp.out
+ env UR_HIP_ENABLE_IMAGE_SUPPORT=1 ONEAPI_DEVICE_SELECTOR=hip:gpu /__w/llvm/llvm/build-e2e/Regression/commandlist/Output/gpu.cpp.tmp.out
malloc_consolidate(): unaligned fastbin chunk detected
/__w/llvm/llvm/build-e2e/Regression/commandlist/Output/gpu.cpp.script: line 1:  1639 Aborted                 (core dumped) env UR_HIP_ENABLE_IMAGE_SUPPORT=1 ONEAPI_DEVICE_SELECTOR=hip:gpu /__w/llvm/llvm/build-e2e/Regression/commandlist/Output/gpu.cpp.tmp.out

--

To reproduce

No response

Environment

Additional context

No response