Fix a crash with std::regex_match during code analysis

[Lablace]

Environment

• OS and Version: macOS Sonoma 14.4.1
• VS Code Version: 1.89.0
• C/C++ Extension Version: v1.19.9
• If using SSH remote, specify OS of remote machine: Ubuntu 22.04.1 (Kernel: 5.15.0)

Bug Summary and Steps to Reproduce

Bug Summary:

When clang-tidy is enabled without tweaking check list, when diagnostic message from it is too long (in below case about 8 wrapped lines in debug output), the extension crashes.

Steps to Reproduce:

• Use the Python code below (at the end section) to generate a main.cpp.
• Open the main.cpp
• Extension crashes.

Expected behavior:

I don't know if vscode have length limitation for diagnostic messages, if not, the extension should work well and display the message intactly, if that's the case, the extension should (at least I think it should) truncate the message and indicate user that the original message is too long to display. Anyway, I think the extension should not crash.

Configuration and Logs

TEMPDIR/cpp-clang-analyzer-optin-padding-mre/main.cpp:1:7: warning: Excessive padding in 'class Sample' (48 padding bytes, where 0 is optimal). Optimal fields order: dvar_000, dvar_001, dvar_002, dvar_003, dvar_004, dvar_005, dvar_006, dvar_007, dvar_008, dvar_009, dvar_00a, dvar_00b, dvar_00c, dvar_00d, dvar_00e, dvar_00f, dvar_010, dvar_011, dvar_012, dvar_013, dvar_014, dvar_015, dvar_016, dvar_017, dvar_018, dvar_019, dvar_01a, dvar_01b, dvar_01c, dvar_01d, dvar_01e, dvar_01f, dvar_020, dvar_021, dvar_022, dvar_023, dvar_024, dvar_025, dvar_026, dvar_027, dvar_028, dvar_029, dvar_02a, dvar_02b, dvar_02c, dvar_02d, dvar_02e, dvar_02f, dvar_030, dvar_031, dvar_032, dvar_033, dvar_034, dvar_035, dvar_036, dvar_037, dvar_038, dvar_039, dvar_03a, dvar_03b, dvar_03c, dvar_03d, dvar_03e, dvar_03f, dvar_040, dvar_041, dvar_042, dvar_043, dvar_044, dvar_045, dvar_046, dvar_047, dvar_048, dvar_049, dvar_04a, dvar_04b, dvar_04c, dvar_04d, dvar_04e, dvar_04f, dvar_050, dvar_051, 
dvar_052, dvar_053, dvar_054, dvar_055, dvar_056, dvar_057, dvar_058, dvar_059, dvar_05a, dvar_05b, dvar_05c, dvar_05d, dvar_05e, dvar_05f, dvar_060, dvar_061, dvar_062, dvar_063, dvar_064, dvar_065, dvar_066, dvar_067, dvar_068, dvar_069, dvar_06a, dvar_06b, dvar_06c, dvar_06d, dvar_06e, dvar_06f, dvar_070, dvar_071, dvar_072, dvar_073, dvar_074, dvar_075, dvar_076, dvar_077, dvar_078, dvar_079, dvar_07a, dvar_07b, dvar_07c, dvar_07d, dvar_07e, dvar_07f, dvar_080, dvar_081, dvar_082, dvar_083, dvar_084, dvar_085, dvar_086, dvar_087, dvar_088, dvar_089, dvar_08a, dvar_08b, dvar_08c, dvar_08d, dvar_08e, dvar_08f, dvar_090, dvar_091, dvar_092, dvar_093, dvar_094, dvar_095, dvar_096, dvar_097, dvar_098, dvar_099, dvar_09a, dvar_09b, dvar_09c, dvar_09d, dvar_09e, dvar_09f, dvar_0a0, dvar_0a1, dvar_0a2, dvar_0a3, dvar_0a4, dvar_0a5, dvar_0a6, dvar_0a7, dvar_0a8, dvar_0a9, dvar_0aa, dvar_0ab, dvar_0ac, dvar_0ad, dvar_0ae, dvar_0af, dvar_0b0, dvar_0b1, dvar_0b2, dvar_0b3, dvar_0b4, dvar_0b5, 
dvar_0b6, dvar_0b7, dvar_0b8, dvar_0b9, dvar_0ba, dvar_0bb, dvar_0bc, dvar_0bd, dvar_0be, dvar_0bf, dvar_0c0, dvar_0c1, dvar_0c2, dvar_0c3, dvar_0c4, dvar_0c5, dvar_0c6, dvar_0c7, dvar_0c8, dvar_0c9, dvar_0ca, dvar_0cb, dvar_0cc, dvar_0cd, dvar_0ce, dvar_0cf, dvar_0d0, dvar_0d1, dvar_0d2, dvar_0d3, dvar_0d4, dvar_0d5, dvar_0d6, dvar_0d7, dvar_0d8, dvar_0d9, dvar_0da, dvar_0db, dvar_0dc, dvar_0dd, dvar_0de, dvar_0df, dvar_0e0, dvar_0e1, dvar_0e2, dvar_0e3, dvar_0e4, dvar_0e5, dvar_0e6, dvar_0e7, dvar_0e8, dvar_0e9, dvar_0ea, dvar_0eb, dvar_0ec, dvar_0ed, dvar_0ee, dvar_0ef, dvar_0f0, dvar_0f1, dvar_0f2, dvar_0f3, dvar_0f4, dvar_0f5, dvar_0f6, dvar_0f7, dvar_0f8, dvar_0f9, dvar_0fa, dvar_0fb, dvar_0fc, dvar_0fd, dvar_0fe, dvar_0ff, dvar_100, dvar_101, dvar_102, dvar_103, dvar_104, dvar_105, dvar_106, dvar_107, dvar_108, dvar_109, dvar_10a, dvar_10b, dvar_10c, dvar_10d, dvar_10e, dvar_10f, dvar_110, dvar_111, dvar_112, dvar_113, dvar_114, dvar_115, dvar_116, dvar_117, dvar_118, dvar_119, 
dvar_11a, dvar_11b, dvar_11c, dvar_11d, dvar_11e, dvar_11f, dvar_120, dvar_121, dvar_122, dvar_123, dvar_124, dvar_125, dvar_126, dvar_127, dvar_128, dvar_129, dvar_12a, dvar_12b, dvar_12c, dvar_12d, dvar_12e, dvar_12f, dvar_130, dvar_131, dvar_132, dvar_133, dvar_134, dvar_135, dvar_136, dvar_137, dvar_138, dvar_139, dvar_13a, dvar_13b, dvar_13c, dvar_13d, dvar_13e, dvar_13f, dvar_140, dvar_141, dvar_142, dvar_143, dvar_144, dvar_145, dvar_146, dvar_147, dvar_148, dvar_149, dvar_14a, dvar_14b, dvar_14c, dvar_14d, dvar_14e, dvar_14f, dvar_150, dvar_151, dvar_152, dvar_153, dvar_154, dvar_155, dvar_156, dvar_157, dvar_158, dvar_159, dvar_15a, dvar_15b, dvar_15c, dvar_15d, dvar_15e, dvar_15f, dvar_160, dvar_161, dvar_162, dvar_163, dvar_164, dvar_165, dvar_166, dvar_167, dvar_168, dvar_169, dvar_16a, dvar_16b, dvar_16c, dvar_16d, dvar_16e, dvar_16f, dvar_170, dvar_171, dvar_172, dvar_173, dvar_174, dvar_175, dvar_176, dvar_177, dvar_178, dvar_179, dvar_17a, dvar_17b, dvar_17c, dvar_17d, 
dvar_17e, dvar_17f, ivar_001_000, ivar_002_000, ivar_002_001, ivar_003_000, ivar_003_001, ivar_003_002, ivar_004_000, ivar_004_001, ivar_004_002, ivar_004_003, ivar_005_000, ivar_005_001, ivar_005_002, ivar_005_003, ivar_005_004, ivar_006_000, ivar_006_001, ivar_006_002, ivar_006_003, ivar_006_004, ivar_006_005, ivar_007_000, ivar_007_001, ivar_007_002, ivar_007_003, ivar_007_004, ivar_007_005, ivar_007_006, ivar_008_000, ivar_008_001, ivar_008_002, ivar_008_003, ivar_008_004, ivar_008_005, ivar_008_006, ivar_008_007, ivar_009_000, ivar_009_001, ivar_009_002, ivar_009_003, ivar_009_004, ivar_009_005, ivar_009_006, ivar_009_007, ivar_009_008, ivar_00a_000, ivar_00a_001, ivar_00a_002, ivar_00a_003, ivar_00a_004, ivar_00a_005, ivar_00a_006, ivar_00a_007, ivar_00a_008, ivar_00a_009, ivar_00b_000, ivar_00b_001, ivar_00b_002, ivar_00b_003, ivar_00b_004, ivar_00b_005, ivar_00b_006, ivar_00b_007, ivar_00b_008, ivar_00b_009, ivar_00b_00a, ivar_00c_000, ivar_00c_001, ivar_00c_002, ivar_00c_003, 
ivar_00c_004, ivar_00c_005, ivar_00c_006, ivar_00c_007, ivar_00c_008, ivar_00c_009, ivar_00c_00a, ivar_00c_00b, ivar_00d_000, ivar_00d_001, ivar_00d_002, ivar_00d_003, ivar_00d_004, ivar_00d_005, ivar_00d_006, ivar_00d_007, ivar_00d_008, ivar_00d_009, ivar_00d_00a, ivar_00d_00b, ivar_00d_00c, ivar_00e_000, ivar_00e_001, ivar_00e_002, ivar_00e_003, ivar_00e_004, ivar_00e_005, ivar_00e_006, ivar_00e_007, ivar_00e_008, ivar_00e_009, ivar_00e_00a, ivar_00e_00b, ivar_00e_00c, ivar_00e_00d, ivar_00f_000, ivar_00f_001, ivar_00f_002, ivar_00f_003, ivar_00f_004, ivar_00f_005, ivar_00f_006, ivar_00f_007, ivar_00f_008, ivar_00f_009, ivar_00f_00a, ivar_00f_00b, ivar_00f_00c, ivar_00f_00d, ivar_00f_00e, ivar_010_000, ivar_010_001, ivar_010_002, ivar_010_003, ivar_010_004, ivar_010_005, ivar_010_006, ivar_010_007, ivar_010_008, ivar_010_009, ivar_010_00a, ivar_010_00b, ivar_010_00c, ivar_010_00d, ivar_010_00e, ivar_010_00f, ivar_011_000, ivar_011_001, ivar_011_002, ivar_011_003, ivar_011_004, 
ivar_011_005, ivar_011_006, ivar_011_007, ivar_011_008, ivar_011_009, ivar_011_00a, ivar_011_00b, ivar_011_00c, ivar_011_00d, ivar_011_00e, ivar_011_00f, ivar_011_010, ivar_012_000, ivar_012_001, ivar_012_002, ivar_012_003, ivar_012_004, ivar_012_005, ivar_012_006, ivar_012_007, ivar_012_008, ivar_012_009, ivar_012_00a, ivar_012_00b, ivar_012_00c, ivar_012_00d, ivar_012_00e, ivar_012_00f, ivar_012_010, ivar_012_011, ivar_013_000, ivar_013_001, ivar_013_002, ivar_013_003, ivar_013_004, ivar_013_005, ivar_013_006, ivar_013_007, ivar_013_008, ivar_013_009, ivar_013_00a, ivar_013_00b, ivar_013_00c, ivar_013_00d, ivar_013_00e, ivar_013_00f, ivar_013_010, ivar_013_011, ivar_013_012, ivar_014_000, ivar_014_001, ivar_014_002, ivar_014_003, ivar_014_004, ivar_014_005, ivar_014_006, ivar_014_007, ivar_014_008, ivar_014_009, ivar_014_00a, ivar_014_00b, ivar_014_00c, ivar_014_00d, ivar_014_00e, ivar_014_00f, ivar_014_010, ivar_014_011, ivar_014_012, ivar_014_013, ivar_015_000, ivar_015_001, 
ivar_015_002, ivar_015_003, ivar_015_004, ivar_015_005, ivar_015_006, ivar_015_007, ivar_015_008, ivar_015_009, ivar_015_00a, ivar_015_00b, ivar_015_00c, ivar_015_00d, ivar_015_00e, ivar_015_00f, ivar_015_010, ivar_015_011, ivar_015_012, ivar_015_013, ivar_015_014, ivar_016_000, ivar_016_001, ivar_016_002, ivar_016_003, ivar_016_004, ivar_016_005, ivar_016_006, ivar_016_007, ivar_016_008, ivar_016_009, ivar_016_00a, ivar_016_00b, ivar_016_00c, ivar_016_00d, ivar_016_00e, ivar_016_00f, ivar_016_010, ivar_016_011, ivar_016_012, ivar_016_013, ivar_016_014, ivar_016_015, ivar_017_000, ivar_017_001, ivar_017_002, ivar_017_003, ivar_017_004, ivar_017_005, ivar_017_006, ivar_017_007, ivar_017_008, ivar_017_009, ivar_017_00a, ivar_017_00b, ivar_017_00c, ivar_017_00d, ivar_017_00e, ivar_017_00f, ivar_017_010, ivar_017_011, ivar_017_012, ivar_017_013, ivar_017_014, ivar_017_015, ivar_017_016, ivar_018_000, ivar_018_001, ivar_018_002, ivar_018_003, ivar_018_004, ivar_018_005, ivar_018_006, 
ivar_018_007, ivar_018_008, ivar_018_009, ivar_018_00a, ivar_018_00b, ivar_018_00c, ivar_018_00d, ivar_018_00e, ivar_018_00f, ivar_018_010, ivar_018_011, ivar_018_012, ivar_018_013, ivar_018_014, ivar_018_015, ivar_018_016, ivar_018_017, consider reordering the fields or adding explicit padding members [clang-analyzer-optin.performance.Padding]

Other Extensions

No response

Additional context

I use this simple Python code to generate a main.cpp, with which clang-tidy applies clang-analyzer-optin.performance.Padding and emits warning: Excessive padding in 'class Sample', along with suggested order (which is too long).

with open("main.cpp", mode="w", encoding="utf8") as f:
    f.write(
        r"class Sample {" + "\n",
    )

    j = 0
    for i in range(384):
        f.write(f"    double dvar_{i:>03x};\n")
        if i % 16 == 0:
            j += 1
            for k in range(j):
                f.write(f"    int ivar_{j:>03x}_{k:>03x} = 0;\n")

    f.write(r"};")

I tried replacing 384 with 256 and this time the extension displayed the diagnostics normally.

I also tried manually disabling the clang-analyzer-optin.performance.Padding check in workspace setting with "C_Cpp.codeAnalysis.clangTidy.checks.disabled": ["clang-analyzer-optin.performance.Padding"], and the extension worked normally too.

[sean-mcmanus]

@Lablace Thanks for reporting this. It repros on Linux but not Windows.

[sean-mcmanus]

@Lablace It's crashing due to a call to std::regex_match with the gcc 11.4 implementation, which seems to rely on some 256 bitset internally:

It doesn't repro with clang/msvc's implementation of std::regex_match so it looks like a gcc bug. In fact, it's fixed if we use gcc 12 instead, but due to our usage of the Linux musl implementation I'm not sure if we can upgrade or not yet.

[Lablace]

Thanks for your detailed clarification, I had never thought this would have something to do with GCC. Let me know if I could offer more help, and feel free to close this if you like or need.

[sean-mcmanus]

@Lablace It's actually a known long-standing issue with the gcc regex implementation that uses the stack too much: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61601 . It's fixed after we increase the thread stack size.

[Lablace]

It just hit my mind that increasing the stack size would solve THIS KIND of problem (like cases when I increase the range to 512 or 1024 in the Python code, though one perhaps should not use a class to store such things) or just the specific case (range is 384)?

[sean-mcmanus]

@Lablace Yeah, you're right. Increasing the stack size isn't really a good fix since it'll crash again with a bigger string. I'll see if I can modify the regex or do some other string manipulation to not make it crash (i.e. limit the size of the matching it's trying to do).

[sean-mcmanus]

@Lablace Fixed with https://github.com/microsoft/vscode-cpptools/releases/tag/v1.21.0 . We changed the regex used so that it wouldn't stack overflow.