[llvm] r339768 - [llvm-mca] Fix PR38575: Avoid an invalid implicit truncation of a processor resource mask (an uint64_t value) to unsigned.
Roman Lebedev via llvm-commits
llvm-commits at lists.llvm.org
Wed Aug 15 06:33:35 PDT 2018
Thanks!
On Wed, Aug 15, 2018 at 3:53 PM, Andrea Di Biagio via llvm-commits
<llvm-commits at lists.llvm.org> wrote:
> Author: adibiagio
> Date: Wed Aug 15 05:53:38 2018
> New Revision: 339768
>
> URL: http://llvm.org/viewvc/llvm-project?rev=339768&view=rev
> Log:
> [llvm-mca] Fix PR38575: Avoid an invalid implicit truncation of a processor resource mask (an uint64_t value) to unsigned.
>
> This patch fixes a regression introduced at revision 338702.
>
> A processor resource mask was incorrectly implicitly truncated to an unsigned
> quantity. Later on, the truncated mask was used to initialize an element of a
> vector of processor resource descriptors.
> On targets with more than 32 processor resources, some elements of the vector
> are left uninitialized. As a consequence, this bug might have eventually caused
> a crash due to null dereference in the Scheduler.
>
> This patch fixes PR38575, and adds a test for it.
>
> Added:
> llvm/trunk/test/tools/llvm-mca/AArch64/Exynos/pr38575.s
> Modified:
> llvm/trunk/tools/llvm-mca/Scheduler.cpp
>
> Added: llvm/trunk/test/tools/llvm-mca/AArch64/Exynos/pr38575.s
> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/tools/llvm-mca/AArch64/Exynos/pr38575.s?rev=339768&view=auto
> ==============================================================================
> --- llvm/trunk/test/tools/llvm-mca/AArch64/Exynos/pr38575.s (added)
> +++ llvm/trunk/test/tools/llvm-mca/AArch64/Exynos/pr38575.s Wed Aug 15 05:53:38 2018
> @@ -0,0 +1,82 @@
> +# NOTE: Assertions have been autogenerated by utils/update_mca_test_checks.py
> +# RUN: llvm-mca -march=aarch64 -mcpu=exynos-m3 < %s | FileCheck %s
> +
> +# Check that llvm-mca doesn't crash when analyzing this block of code.
> +
> +# This test is related to PR38575.
> +# The crash was indirectly cause by an incorrectly truncation of a processor
> +# resource mask value (an uint64_t) to int value.
> +# The problem can only be observed on targets that declare more than 32
> +# resources (like exynos-m3).
> +
> +ror x1, x2, x3
> +
> +# CHECK: Iterations: 100
> +# CHECK-NEXT: Instructions: 100
> +# CHECK-NEXT: Total Cycles: 28
> +# CHECK-NEXT: Dispatch Width: 6
> +# CHECK-NEXT: IPC: 3.57
> +# CHECK-NEXT: Block RThroughput: 0.3
> +
> +# CHECK: Instruction Info:
> +# CHECK-NEXT: [1]: #uOps
> +# CHECK-NEXT: [2]: Latency
> +# CHECK-NEXT: [3]: RThroughput
> +# CHECK-NEXT: [4]: MayLoad
> +# CHECK-NEXT: [5]: MayStore
> +# CHECK-NEXT: [6]: HasSideEffects (U)
> +
> +# CHECK: [1] [2] [3] [4] [5] [6] Instructions:
> +# CHECK-NEXT: 1 1 0.25 ror x1, x2, x3
> +
> +# CHECK: Resources:
> +# CHECK-NEXT: [0] - M3PipeF0
> +# CHECK-NEXT: [1] - M3PipeF1
> +# CHECK-NEXT: [2] - M3PipeF2
> +# CHECK-NEXT: [3.0] - M3UnitA
> +# CHECK-NEXT: [3.1] - M3UnitA
> +# CHECK-NEXT: [4.0] - M3UnitB
> +# CHECK-NEXT: [4.1] - M3UnitB
> +# CHECK-NEXT: [5.0] - M3UnitC
> +# CHECK-NEXT: [5.1] - M3UnitC
> +# CHECK-NEXT: [6] - M3UnitD
> +# CHECK-NEXT: [7] - M3UnitFADD0
> +# CHECK-NEXT: [8] - M3UnitFADD1
> +# CHECK-NEXT: [9] - M3UnitFADD2
> +# CHECK-NEXT: [10] - M3UnitFCVT0
> +# CHECK-NEXT: [11] - M3UnitFCVT1
> +# CHECK-NEXT: [12.0] - M3UnitFDIV0
> +# CHECK-NEXT: [12.1] - M3UnitFDIV0
> +# CHECK-NEXT: [13.0] - M3UnitFDIV1
> +# CHECK-NEXT: [13.1] - M3UnitFDIV1
> +# CHECK-NEXT: [14] - M3UnitFMAC0
> +# CHECK-NEXT: [15] - M3UnitFMAC1
> +# CHECK-NEXT: [16] - M3UnitFMAC2
> +# CHECK-NEXT: [17.0] - M3UnitFSQR
> +# CHECK-NEXT: [17.1] - M3UnitFSQR
> +# CHECK-NEXT: [18] - M3UnitFST0
> +# CHECK-NEXT: [19] - M3UnitFST1
> +# CHECK-NEXT: [20.0] - M3UnitL
> +# CHECK-NEXT: [20.1] - M3UnitL
> +# CHECK-NEXT: [21] - M3UnitNALU0
> +# CHECK-NEXT: [22] - M3UnitNALU1
> +# CHECK-NEXT: [23] - M3UnitNALU2
> +# CHECK-NEXT: [24] - M3UnitNCRY0
> +# CHECK-NEXT: [25] - M3UnitNCRY1
> +# CHECK-NEXT: [26] - M3UnitNMSC
> +# CHECK-NEXT: [27] - M3UnitNMUL
> +# CHECK-NEXT: [28] - M3UnitNSHF0
> +# CHECK-NEXT: [29] - M3UnitNSHF1
> +# CHECK-NEXT: [30] - M3UnitNSHF2
> +# CHECK-NEXT: [31] - M3UnitNSHT0
> +# CHECK-NEXT: [32] - M3UnitNSHT1
> +# CHECK-NEXT: [33] - M3UnitNSHT2
> +# CHECK-NEXT: [34] - M3UnitS
> +
> +# CHECK: Resource pressure per iteration:
> +# CHECK-NEXT: [0] [1] [2] [3.0] [3.1] [4.0] [4.1] [5.0] [5.1] [6] [7] [8] [9] [10] [11] [12.0] [12.1] [13.0] [13.1] [14] [15] [16] [17.0] [17.1] [18] [19] [20.0] [20.1] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34]
> +# CHECK-NEXT: - - - 0.25 0.25 - - 0.25 0.25 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
> +
> +# CHECK: Resource pressure by instruction:
> +# CHECK-NEXT: [0] [1] [2] [3.0] [3.1] [4.0] [4.1] [5.0] [5.1] [6] [7] [8] [9] [10] [11] [12.0] [12.1] [13.0] [13.1] [14] [15] [16] [17.0] [17.1] [18] [19] [20.0] [20.1] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] Instructions:
> +# CHECK-NEXT: - - - 0.25 0.25 - - 0.25 0.25 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ror x1, x2, x3
>
> Modified: llvm/trunk/tools/llvm-mca/Scheduler.cpp
> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/tools/llvm-mca/Scheduler.cpp?rev=339768&r1=339767&r2=339768&view=diff
> ==============================================================================
> --- llvm/trunk/tools/llvm-mca/Scheduler.cpp (original)
> +++ llvm/trunk/tools/llvm-mca/Scheduler.cpp Wed Aug 15 05:53:38 2018
> @@ -59,7 +59,7 @@ void ResourceManager::initialize(const l
> Resources.resize(SM.getNumProcResourceKinds());
>
> for (unsigned I = 0, E = SM.getNumProcResourceKinds(); I < E; ++I) {
> - unsigned Mask = ProcResID2Mask[I];
> + uint64_t Mask = ProcResID2Mask[I];
> Resources[getResourceStateIndex(Mask)] =
> llvm::make_unique<ResourceState>(*SM.getProcResource(I), I, Mask);
> }
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits
More information about the llvm-commits
mailing list