Support runtime SIMD #1
Assignees
Comments
jiaoew1991
pushed a commit
to jiaoew1991/folly
that referenced
this issue
Nov 2, 2023
Summary:
AsyncUDPSocket test cases are crashing when running under llvm15. During
debugging it seems that the issue is the fact that the code tries to allocated 0
size array. Changing the code to prevent such allocation.
This is not very clean why to fix, but I am not sure if there is better one.
Please let me know if you have any suggestions.
Sample crash:
```
$ buck test //folly/io/async/test:async_udp_socket_test
...
stdout:
Note: Google Test filter = AsyncUDPSocketTest.TestDetachAttach
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from AsyncUDPSocketTest
[ RUN ] AsyncUDPSocketTest.TestDetachAttach
stderr:
fbcode/folly/io/async/AsyncUDPSocket.cpp:699:10: runtime error: variable length array bound evaluates to non-positive value 0
#0 0x7f4d8ed93704 in folly::AsyncUDPSocket::writev(folly::SocketAddress const&, iovec const*, unsigned long, int) fbcode/folly/io/async/AsyncUDPSocket.cpp:698
milvus-io#1 0x7f4d8ed9081f in folly::AsyncUDPSocket::writeGSO(folly::SocketAddress const&, std::unique_ptr<folly::IOBuf, std::default_delete<folly::IOBuf>> const&, int) fbcode/folly/io/async/AsyncUDPSocket.cpp:528
milvus-io#2 0x7f4d8ed900b2 in folly::AsyncUDPSocket::write(folly::SocketAddress const&, std::unique_ptr<folly::IOBuf, std::default_delete<folly::IOBuf>> const&) fbcode/folly/io/async/AsyncUDPSocket.cpp:660
milvus-io#3 0x350a05 in AsyncUDPSocketTest_TestDetachAttach_Test::TestBody() fbcode/folly/io/async/test/AsyncUDPSocketTest.cpp:914
milvus-io#4 0x7f4d90dd1ad5 in testing::Test::Run() /home/engshare/third-party2/googletest/1.11.0/src/googletest/googletest/src/gtest.cc:2682:50
milvus-io#5 0x7f4d90dd1ad5 in testing::Test::Run() /home/engshare/third-party2/googletest/1.11.0/src/googletest/googletest/src/gtest.cc:2672:6
milvus-io#6 0x7f4d90dd1c64 in testing::TestInfo::Run() /home/engshare/third-party2/googletest/1.11.0/src/googletest/googletest/src/gtest.cc:2861:14
milvus-io#7 0x7f4d90dd1c64 in testing::TestInfo::Run() /home/engshare/third-party2/googletest/1.11.0/src/googletest/googletest/src/gtest.cc:2833:6
milvus-io#8 0x7f4d90dd2321 in testing::TestSuite::Run() /home/engshare/third-party2/googletest/1.11.0/src/googletest/googletest/src/gtest.cc:3015:31
milvus-io#9 0x7f4d90dd2321 in testing::TestSuite::Run() /home/engshare/third-party2/googletest/1.11.0/src/googletest/googletest/src/gtest.cc:2993:6
milvus-io#10 0x7f4d90dd2b1e in testing::internal::UnitTestImpl::RunAllTests() /home/engshare/third-party2/googletest/1.11.0/src/googletest/googletest/src/gtest.cc:5855:47
milvus-io#11 0x7f4d90dd1d87 in bool testing::internal::HandleExceptionsInMethodIfSupported<testing::internal::UnitTestImpl, bool>(testing::internal::UnitTestImpl*, bool (testing::internal::UnitTestImpl::*)(), char const*) /home/engshare/third-party2/googletest/1.11.0/src/googletest/googletest/src/gtest.cc:2665:29
milvus-io#12 0x7f4d90dd1d87 in testing::UnitTest::Run() /home/engshare/third-party2/googletest/1.11.0/src/googletest/googletest/src/gtest.cc:5438:55
milvus-io#13 0x7f4d90d5c990 in RUN_ALL_TESTS() fbcode/third-party-buck/platform010/build/googletest/include/gtest/gtest.h:2490
milvus-io#14 0x7f4d90d5c618 in main fbcode/common/gtest/LightMain.cpp:20
milvus-io#15 0x7f4d8ea2c656 in __libc_start_call_main /home/engshare/third-party2/glibc/2.34/src/glibc-2.34/csu/../sysdeps/nptl/libc_start_call_main.h:58:16
milvus-io#16 0x7f4d8ea2c717 in __libc_start_main@GLIBC_2.2.5 /home/engshare/third-party2/glibc/2.34/src/glibc-2.34/csu/../csu/libc-start.c:409:3
milvus-io#17 0x33ea60 in _start /home/engshare/third-party2/glibc/2.34/src/glibc-2.34/csu/../sysdeps/x86_64/start.S:116
...
```
Reviewed By: russoue, dmm-fb
Differential Revision: D43858875
fbshipit-source-id: 93749bab17027b6dfc0dbc01b6c183e501a5494c
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
F14Table and F14Map have 2 versions, one for sse4.2 and neon, and another for other arches.
We do not want to maintain different binary for different arches, so we should support runtime SIMD adaptor to simplify it.
The text was updated successfully, but these errors were encountered: