mintty crashes on Windows 7

mintty crashes on Windows 7

[Orgad Shaneh]

[-- Attachment #1: Type: text/plain, Size: 1243 bytes --]

Hi,

Running `mintty ./bash` crashes on Windows 7 on cygwin-3_3-branch.

Tested in MSYS2 on merge-3.3 branch from
https://github.com/orgads/msys2-runtime-1. It includes the tip of
cygwin-3_3-branch as of today (05827d2df8).

Steps to reproduce:
1. Start cmd.exe
2. cd <msys-directory>\usr\bin
3. mintty ./bash

GDB trace attached. Simplified form:

1  ntdll!KiRaiseUserExceptionDispatcher                       0x7791b5ba
2  KERNELBASE!CloseHandle                                     0x7feffb31863
3  KERNEL32!CloseHandle                                       0x776b14f1
4  fhandler_base::close                 fhandler.cc      1217 0x1800682db
5  fhandler_pipe::close                 fhandler_pipe.cc 685  0x18009069a
6  fhandler_base::close_with_arch       fhandler.cc      1193 0x18006c7d5
7  close_all_files                      syscalls.cc      102  0x18014548f
8  do_exit                              dcrt0.cc         1200 0x180048213
9  _exit                                dcrt0.cc         1317 0x1800483ef
10 exit                                 exit.c           64   0x1801b3bdd
11 cygwin_exit                          dcrt0.cc         1311 0x1800483d3
12 _sigfe                               sigfe.s          35   0x18019593b
13 ??

[-- Attachment #2: gdb-trace.txt --]
[-- Type: text/plain, Size: 16597 bytes --]


Thread 5 (Thread 6896.0x1eb8 "waitproc"):
#0  0x000000007791980a in ntdll!ZwReadFile () from C:\Windows\SYSTEM32\ntdll.dll
No symbol table info available.
#1  0x000007feffb31a6a in ReadFile () from C:\Windows\system32\KernelBase.dll
No symbol table info available.
#2  0x00000000776a0599 in ReadFile () from C:\Windows\system32\kernel32.dll
No symbol table info available.
#3  0x000000018010ca04 in proc_waiter (arg=<optimized out>) at ../../../../winsup/cygwin/pinfo.cc:1214
        nb = 0
        buf = 0 '\000'
        vchild = {<pinfo_minimal> = {h = 0x228, hProcess = 0x22c, rd_proc_pipe = 0x220}, destroy = false, winpid_hdl = 0x0, procinfo = 0x3a0000, waiter_ready = false, wait_thread = 0x180238898 <threads+88>}
        si = {si_signo = 20, si_code = 28, si_pid = 502, si_uid = 0, si_errno = 0, {__pad = {0 <repeats 32 times>}, _si_commune = {_si_code = 0, _si_read_handle = 0x0, _si_write_handle = 0x0, _si_process_handle = 0x0, {{_si_fd = 0, _si_flags = 0}, _si_pipe_unique_id = 0, _si_str = 0x0}}, {{si_sigval = {sival_int = 0, sival_ptr = 0x0}, si_value = {sival_int = 0, sival_ptr = 0x0}}, {si_tid = 0, si_overrun = 0}}, {si_status = 0, si_utime = 0, si_stime = 0}, si_addr = 0x0, {__pad2 = {0 <repeats 30 times>}, si_cyg = 0x0}}}
        pid = 502
        its_me = false
        __PRETTY_FUNCTION__ = "DWORD proc_waiter(void*)"
#4  0x0000000180046583 in cygthread::callfunc (this=this@entry=0x180238898 <threads+88>, issimplestub=issimplestub@entry=false) at ../../../../winsup/cygwin/cygthread.cc:48
        pass_arg = <optimized out>
#5  0x0000000180046b46 in cygthread::stub (arg=arg@entry=0x180238898 <threads+88>) at ../../../../winsup/cygwin/cygthread.cc:91
        notify = <optimized out>
        info = 0x180238898 <threads+88>
        __PRETTY_FUNCTION__ = "static DWORD cygthread::stub(void*)"
#6  0x0000000180047716 in _cygtls::call2 (this=0x24ace00, func=0x180046ac0 <cygthread::stub(void*)>, arg=0x180238898 <threads+88>, buf=buf@entry=0x24acd50) at ../../../../winsup/cygwin/cygtls.cc:40
        res = <optimized out>
#7  0x00000001800477c4 in _cygtls::call (func=<optimized out>, arg=<optimized out>) at ../../../../winsup/cygwin/cygtls.cc:27
        buf = '\000' <repeats 2360 times>, "?M#\200\001\000\000\000\220N#\200\001\000\000\000HO#\200\001", '\000' <repeats 51 times>, "\377\377\377\377\000\000\000\000\220Y\033\200\001", '\000' <repeats 139 times>, "\001\000\000\000\000\000\000\000\016\063\315?4\022m?\336?\005\000\v", '\000' <repeats 1067 times>, "\003\000\000\000\000\000\000\000?M#\200\001", '\000' <repeats 587 times>, "\200\326?J\002", '\000' <repeats 28 times>, "\002", '\000' <repeats 1343 times>, "?\036", '\000' <repeats 158 times>, "\230\210#\200\001", '\000' <repeats 75 times>, "X?J\002", '\000' <repeats 2052 times>, "?\027c\307", '\000' <repeats 92 times>, "\nTjw", '\000' <repeats 1388 times>, "\037\000\004\033\000\000\000\000\020\004\000\000\000\000\000\000\210\371?<\000\000\000\000\000\001\000\000\000\000\000\000\000\000\000;\000\000\000\000\000k\001,P\000\000\000\000??\230w\000\000\000\000\212?<\000\000\000\000\000??<\000\000\000\000\000\200\371?<", '\000' <repeats 13 times>, "\001\000\000\000\000\000\000\000D\000\000\000\000\000\000\000 #", '\000' <repeats 30 times>, "\004\000\000\000\003\300\002\005\000\000\000\000\000\000\000\000\200\371?<\000\000\000\000\000\220?<\000\000\000\000\000\177\000\000\000\000\000\000\000\220?<", '\000' <repeats 13 times>, "D", '\000' <repeats 11 times>, "\177\000\000\000 #", '\000' <repeats 14 times>, "D", '\000' <repeats 15 times>, "\020\004", '\000' <repeats 14 times>, "\220?<\000\000\000\000\000\060\000\000\000\000\000\000\000\060\002;", '\000' <repeats 13 times>, " 6\002", '\000' <repeats 13 times>, "t\002;\000\000\000\000\000\060\002;\000\000\000\000\000\177", '\000' <repeats 39 times>, "\060\002;", '\000' <repeats 45 times>, "d#\004C", '\000' <repeats 28 times>, "\002\000\004\006", '\000' <repeats 28 times>, "(1?w", '\000' <repeats 22 times>, ";\000\000\000\000\000\000\000;\000\000\000\000\000\020\004", '\000' <repeats 14 times>, "\234{\215w\000\000\000\000\000\000;\000\000\000\000\000k\001,P\000\000\000\000\020\004\000\000\000\000\000\000@\004", '\000' <repeats 14 times>...
        protect = <optimized out>
#8  0x00000000776a556d in KERNEL32!BaseThreadInitThunk () from C:\Windows\system32\kernel32.dll
No symbol table info available.
#9  0x000000007790372d in ntdll!RtlUserThreadStart () from C:\Windows\SYSTEM32\ntdll.dll
No symbol table info available.
#10 0x0000000000000000 in ?? ()
No symbol table info available.
Backtrace stopped: previous frame inner to this frame (corrupt stack?)

Thread 4 (Thread 6896.0x187c):
#0  0x000000007791b0aa in ntdll!ZwWaitForWorkViaWorkerFactory () from C:\Windows\SYSTEM32\ntdll.dll
No symbol table info available.
#1  0x00000000779a97c4 in ntdll!DbgUiRemoteBreakin () from C:\Windows\SYSTEM32\ntdll.dll
No symbol table info available.
#2  0x0000000180047716 in _cygtls::call2 (this=0x22ace00, func=0x779a94c0 <ntdll!DbgUiRemoteBreakin+80>, arg=0x3c8d00, buf=buf@entry=0x22acd50) at ../../../../winsup/cygwin/cygtls.cc:40
        res = <optimized out>
#3  0x00000001800477c4 in _cygtls::call (func=<optimized out>, arg=<optimized out>) at ../../../../winsup/cygwin/cygtls.cc:27
        buf = '\000' <repeats 2360 times>, "?M#\200\001\000\000\000\220N#\200\001\000\000\000HO#\200\001", '\000' <repeats 203 times>, "\001\000\000\000\000\000\000\000\016\063\315?4\022m?\336?\005\000\v", '\000' <repeats 1067 times>, "\003\000\000\000\000\000\000\000?M#\200\001", '\000' <repeats 587 times>, "\200\326?*\002", '\000' <repeats 28 times>, "\002", '\000' <repeats 1343 times>, "|\030", '\000' <repeats 238 times>, "X?*\002", '\000' <repeats 2052 times>, "?\027c\307", '\000' <repeats 92 times>, "\nTjw", '\000' <repeats 1108 times>, "?o_\000\000\000\000\000?\214?w", '\000' <repeats 12 times>, "?o_\000\000\000\000\000\300__", '\000' <repeats 29 times>, "X\001_\000\000\000\000\000?\214?w\000\000\000\000\300o_", '\000' <repeats 45 times>, "x?*\002\000\000\000\000\004\000\001\005\000\000\000\000\000\000_\000\000\000\000\000?__\000\000\000\000\000\001\000\000\000\000\000\000\000?\002\000\000\000\000\000\000k\001\000P\000\000\000\000\061\000\000\000\000\000\000\000?\\_\000\000\000\000\000\020\003\000\000\000\000\000\000k\001\000P\000\000\000\000\063\020\230w\000\000\000\000\000\000_\000\000\000\000\000X\001_\000\000\000\000\000\000?*\002\000\000\000\000/\001", '\000' <repeats 30 times>, "\004\000\004\000\000\000\000\000\020\004\000\000\000\000\000\000X\303<\000\000\000\000\000\001\000\000\000\000\000\000\000\000\000;\000\000\000\000\000k\001,P\000\000\000\000??\230w\000\000\000\000Z\303<\000\000\000\000\000[\303<\000\000\000\000\000P\303<", '\000' <repeats 13 times>, "\001\000\000\000\000\000\000\000D\000\000\000\000\000\000\000?", '\000' <repeats 31 times>, "\004\000\000\000\003\300\002\005\000\000\000\000\000\000\000\000P\303<\000\000\000\000\000`\303<\000\000\000\000\000\177\000\000\000\000\000\000\000`\303<", '\000' <repeats 13 times>, "D\000\000\000\000\000\000\000\177\000\000\000\177\000\000\000?", '\000' <repeats 15 times>, "D", '\000' <repeats 15 times>, "\020\004\000\000\000\000\000\000\004\000\000\000\003\000\000\005`\303<\000\000\000\000\000\060\000\000\000\000\000\000\000\060\002;\000\000\000\000\000\177\000\000\000\000\000\000\000P\f", '\000' <repeats 14 times>, "t\002;\000\000\000\000\000\060\002;\000\000\000\000\000\177", '\000' <repeats 15 times>, "\006", '\000' <repeats 15 times>, "\060\002_\000\000\000\000\000\060\002;\000\000\000\000\000`\303<\000\000\000\000\000\070\000\000\000\000\000\000\000\060\002;", '\000' <repeats 13 times>, "@\000\000\000\000\000\000\000\307\000\004\303\000\000\000\000h\002;\000\000\000\000\000\060\002;\000\000\000\000\000\006\000\000\000\000\000\000\000\002\000\004\006", '\000' <repeats 28 times>, "(1?w", '\000' <repeats 22 times>, ";\000\000\000\000\000\000\000;\000\000\000\000\000\020\004", '\000' <repeats 14 times>, "\234{\215w\000\000\000\000\000\000;\000\000\000\000\000k\001,P\000\000\000\000\020\004\000\000\000\000\000\000@\004", '\000' <repeats 14 times>...
        protect = <optimized out>
#4  0x00000000776a556d in KERNEL32!BaseThreadInitThunk () from C:\Windows\system32\kernel32.dll
No symbol table info available.
#5  0x000000007790372d in ntdll!RtlUserThreadStart () from C:\Windows\SYSTEM32\ntdll.dll
No symbol table info available.
#6  0x0000000000000000 in ?? ()
No symbol table info available.
Backtrace stopped: previous frame inner to this frame (corrupt stack?)

Thread 3 (Thread 6896.0xef0):
#0  0x0000000077919d5a in ntdll!ZwWaitForMultipleObjects () from C:\Windows\SYSTEM32\ntdll.dll
No symbol table info available.
#1  0x00000000778c332d in ntdll!RtlIsCriticalSectionLockedByThread () from C:\Windows\SYSTEM32\ntdll.dll
No symbol table info available.
#2  0x00000000776a556d in KERNEL32!BaseThreadInitThunk () from C:\Windows\system32\kernel32.dll
No symbol table info available.
#3  0x000000007790372d in ntdll!RtlUserThreadStart () from C:\Windows\SYSTEM32\ntdll.dll
No symbol table info available.
#4  0x0000000000000000 in ?? ()
No symbol table info available.
Backtrace stopped: previous frame inner to this frame (corrupt stack?)

Thread 2 (Thread 6896.0x80c "sig"):
#0  0x000000007791980a in ntdll!ZwReadFile () from C:\Windows\SYSTEM32\ntdll.dll
No symbol table info available.
#1  0x000007feffb31a6a in ReadFile () from C:\Windows\system32\KernelBase.dll
No symbol table info available.
#2  0x00000000776a0599 in ReadFile () from C:\Windows\system32\kernel32.dll
No symbol table info available.
#3  0x00000001801356c1 in wait_sig () at ../../../../winsup/cygwin/sigproc.cc:1359
        nb = 0
        q = <optimized out>
        pack = {si = {si_signo = 0, si_code = 0, si_pid = 0, si_uid = 0, si_errno = 0, {__pad = {0 <repeats 32 times>}, _si_commune = {_si_code = 0, _si_read_handle = 0x0, _si_write_handle = 0x0, _si_process_handle = 0x0, {{_si_fd = 0, _si_flags = 0}, _si_pipe_unique_id = 0, _si_str = 0x0}}, {{si_sigval = {sival_int = 0, sival_ptr = 0x0}, si_value = {sival_int = 0, sival_ptr = 0x0}}, {si_tid = 0, si_overrun = 0}}, {si_status = 0, si_utime = 0, si_stime = 0}, si_addr = 0x0, {__pad2 = {0 <repeats 30 times>}, si_cyg = 0x0}}}, pid = 0, sigtls = 0x0, mask = 0x0, {wakeup = 0x0, thread_handle = 0x0, next = 0x0}}
        dummy_mask = 0
        tl_entry = <optimized out>
        clearwait = <optimized out>
        sig_held = false
        __PRETTY_FUNCTION__ = "void wait_sig(void*)"
#4  0x0000000180046583 in cygthread::callfunc (this=this@entry=0x180238840 <threads>, issimplestub=issimplestub@entry=false) at ../../../../winsup/cygwin/cygthread.cc:48
        pass_arg = <optimized out>
#5  0x0000000180046b46 in cygthread::stub (arg=arg@entry=0x180238840 <threads>) at ../../../../winsup/cygwin/cygthread.cc:91
        notify = <optimized out>
        info = 0x180238840 <threads>
        __PRETTY_FUNCTION__ = "static DWORD cygthread::stub(void*)"
#6  0x0000000180047716 in _cygtls::call2 (this=0x20ace00, func=0x180046ac0 <cygthread::stub(void*)>, arg=0x180238840 <threads>, buf=buf@entry=0x20acd50) at ../../../../winsup/cygwin/cygtls.cc:40
        res = <optimized out>
#7  0x00000001800477c4 in _cygtls::call (func=<optimized out>, arg=<optimized out>) at ../../../../winsup/cygwin/cygtls.cc:27
        buf = '\000' <repeats 2360 times>, "?M#\200\001\000\000\000\220N#\200\001\000\000\000HO#\200\001", '\000' <repeats 203 times>, "\001\000\000\000\000\000\000\000\016\063\315?4\022m?\336?\005\000\v", '\000' <repeats 1067 times>, "\003\000\000\000\000\000\000\000?M#\200\001", '\000' <repeats 587 times>, "\200\326?\n\002", '\000' <repeats 28 times>, "\002", '\000' <repeats 1343 times>, "\f\b", '\000' <repeats 158 times>, "@\210#\200\001", '\000' <repeats 75 times>, "X?\n\002", '\000' <repeats 2052 times>, "?\027c\307", '\000' <repeats 92 times>, "\nTjw", '\000' <repeats 1108 times>, "?_<\000\000\000\000\000?\214?w", '\000' <repeats 12 times>, "?_<\000\000\000\000\000\300O<", '\000' <repeats 29 times>, "X\001;\000\000\000\000\000?\214?w\000\000\000\000\a\000\004\003", '\000' <repeats 44 times>, "x?\n\002\000\000\000\000\004\000\001\005\000\000\000\000\000\000;\000\000\000\000\000?O<\000\000\000\000\000\001\000\000\000\000\000\000\000\020\004\000\000\000\000\000\000k\001,P\000\000\000\000D\000\000\000\000\000\000\000\220M<\000\000\000\000\000@\004\000\000\000\000\000\000k\001,P\000\000\000\000\063\020\230w\000\000\000\000\000\000;\000\000\000\000\000X\001;\000\000\000\000\000\000?\n\002\000\000\000\000#\001", '\000' <repeats 30 times>, "\a\000\004\003\000\000\000\000\020\004\000\000\000\000\000\000\230M<\000\000\000\000\000\001\000\000\000\000\000\000\000\000\000;\000\000\000\000\000k\001,P\000\000\000\000??\230w\000\000\000\000\232M<\000\000\000\000\000?M<\000\000\000\000\000\220M<", '\000' <repeats 13 times>, "\001\000\000\000\000\000\000\000D\000\000\000\000\000\000\000\337", '\000' <repeats 31 times>, "\004\000\000\000\003\300\002\005\000\000\000\000\000\000\000\000\220M<\000\000\000\000\000X\001;", '\000' <repeats 13 times>, "?M<", '\000' <repeats 13 times>, "D\000\000\000\000\000\000\000\177\000\000\000\177\000\000\000\337", '\000' <repeats 15 times>, "D", '\000' <repeats 15 times>, "\020\004", '\000' <repeats 14 times>, "?M<\000\000\000\000\000\060\000\000\000\000\000\000\000\060\002;\000\000\000\000\000\177\000\000\000\000\000\000\000\020\022", '\000' <repeats 22 times>, "\060\002;\000\000\000\000\000\177", '\000' <repeats 31 times>, "\060\002;\000\000\000\000\000\060\002_", '\000' <repeats 45 times>, "#\000\004'", '\000' <repeats 28 times>, "?\000\004?", '\000' <repeats 28 times>, "(1?w", '\000' <repeats 22 times>, ";\000\000\000\000\000\000\000;\000\000\000\000\000\020\004", '\000' <repeats 14 times>, "\234{\215w\000\000\000\000\000\000;\000\000\000\000\000k\001,P\000\000\000\000\020\004\000\000\000\000\000\000@\004", '\000' <repeats 14 times>...
        protect = <optimized out>
#8  0x00000000776a556d in KERNEL32!BaseThreadInitThunk () from C:\Windows\system32\kernel32.dll
No symbol table info available.
#9  0x000000007790372d in ntdll!RtlUserThreadStart () from C:\Windows\SYSTEM32\ntdll.dll
No symbol table info available.
#10 0x0000000000000000 in ?? ()
No symbol table info available.
Backtrace stopped: previous frame inner to this frame (corrupt stack?)

Thread 1 (Thread 6896.0x12f0 "mintty"):
#0  0x000000007791b5ba in ntdll!KiRaiseUserExceptionDispatcher () from C:\Windows\SYSTEM32\ntdll.dll
No symbol table info available.
#1  0x000007feffb31863 in KERNELBASE!CloseHandle () from C:\Windows\system32\KernelBase.dll
No symbol table info available.
#2  0x00000000776b14f1 in KERNEL32!CloseHandle () from C:\Windows\system32\kernel32.dll
No symbol table info available.
#3  0x00000001800682db in fhandler_base::close (this=this@entry=0x180353e80) at ../../../../winsup/cygwin/fhandler.cc:1217
        res = -1
        __PRETTY_FUNCTION__ = "virtual int fhandler_base::close()"
#4  0x000000018009069a in fhandler_pipe::close (this=0x180353e80) at ../../../../winsup/cygwin/fhandler_pipe.cc:685
        ret = <optimized out>
#5  0x000000018006c7d5 in fhandler_base::close_with_arch (this=0x180353e80) at ../../../../winsup/cygwin/fhandler.cc:1193
        res = <optimized out>
        fh = 0x180353e80
        __PRETTY_FUNCTION__ = "int fhandler_base::close_with_arch()"
#6  0x000000018014548f in close_all_files (norelease=norelease@entry=false) at ../../../../winsup/cygwin/syscalls.cc:102
        cfd = {<cygheap_fdmanip> = {_vptr.cygheap_fdmanip = <optimized out>, fd = 1, locked = false}, fh = 0x180353e80}
        i = 1
        h = 0x0
#7  0x0000000180048213 in do_exit (status=0) at ../../../../winsup/cygwin/dcrt0.cc:1200
        __PRETTY_FUNCTION__ = "void do_exit(int)"
        until_exit = {skip_unlock = true, static locker = {name = 0x1802714f6 <msg1+4086> "lock_process", sync = 1, waiters = 0, bruteforce = 0x20, visits = 2, tls = 0xffffce00}}
        n = <optimized out>
#8  0x00000001800483ef in _exit (n=<optimized out>) at ../../../../winsup/cygwin/dcrt0.cc:1317
No locals.
#9  0x00000001801b3bdd in exit (code=0) at ../../../../../newlib/libc/stdlib/exit.c:64
No locals.
#10 0x00000001800483d3 in cygwin_exit (n=-17808) at ../../../../winsup/cygwin/dcrt0.cc:1311
No locals.
#11 0x000000018019593b in _sigfe () at sigfe.s:35
No locals.
#12 0x0000000000000000 in ?? ()
No symbol table info available.
Backtrace stopped: previous frame inner to this frame (corrupt stack?)

Re: mintty crashes on Windows 7

[Takashi Yano]

On Tue, 3 May 2022 14:47:17 +0300
Orgad Shaneh wrote:
> Hi,
> 
> Running `mintty ./bash` crashes on Windows 7 on cygwin-3_3-branch.
> 
> Tested in MSYS2 on merge-3.3 branch from
> https://github.com/orgads/msys2-runtime-1. It includes the tip of
> cygwin-3_3-branch as of today (05827d2df8).
> 
> Steps to reproduce:
> 1. Start cmd.exe
> 2. cd <msys-directory>\usr\bin
> 3. mintty ./bash
> 
> GDB trace attached. Simplified form:
> 
> 1  ntdll!KiRaiseUserExceptionDispatcher                       0x7791b5ba
> 2  KERNELBASE!CloseHandle                                     0x7feffb31863
> 3  KERNEL32!CloseHandle                                       0x776b14f1
> 4  fhandler_base::close                 fhandler.cc      1217 0x1800682db
> 5  fhandler_pipe::close                 fhandler_pipe.cc 685  0x18009069a
> 6  fhandler_base::close_with_arch       fhandler.cc      1193 0x18006c7d5
> 7  close_all_files                      syscalls.cc      102  0x18014548f
> 8  do_exit                              dcrt0.cc         1200 0x180048213
> 9  _exit                                dcrt0.cc         1317 0x1800483ef
> 10 exit                                 exit.c           64   0x1801b3bdd
> 11 cygwin_exit                          dcrt0.cc         1311 0x1800483d3
> 12 _sigfe                               sigfe.s          35   0x18019593b
> 13 ??

I cannot reproduce this with cygwin git head of cygwin-3_3-branch.
Is this MSYS2 specific problem?

-- 
Takashi Yano <takashi.yano@nifty.ne.jp>

Re: mintty crashes on Windows 7

[Orgad Shaneh]

On Tue, May 3, 2022 at 6:23 PM Takashi Yano <takashi.yano@nifty.ne.jp> wrote:
>
> On Tue, 3 May 2022 14:47:17 +0300
> Orgad Shaneh wrote:
> > Hi,
> >
> > Running `mintty ./bash` crashes on Windows 7 on cygwin-3_3-branch.
> >
> > Tested in MSYS2 on merge-3.3 branch from
> > https://github.com/orgads/msys2-runtime-1. It includes the tip of
> > cygwin-3_3-branch as of today (05827d2df8).
> >
> I cannot reproduce this with cygwin git head of cygwin-3_3-branch.
> Is this MSYS2 specific problem?

You're right. I can't reproduce either. Only in my MSYS branch.

Can you give me some guidance how to debug it?

- Orgad

Re: mintty crashes on Windows 7

[Takashi Yano]

On Tue, 3 May 2022 18:52:28 +0300
Orgad Shaneh wrote:
> On Tue, May 3, 2022 at 6:23 PM Takashi Yano <takashi.yano@nifty.ne.jp> wrote:
> >
> > On Tue, 3 May 2022 14:47:17 +0300
> > Orgad Shaneh wrote:
> > > Hi,
> > >
> > > Running `mintty ./bash` crashes on Windows 7 on cygwin-3_3-branch.
> > >
> > > Tested in MSYS2 on merge-3.3 branch from
> > > https://github.com/orgads/msys2-runtime-1. It includes the tip of
> > > cygwin-3_3-branch as of today (05827d2df8).
> > >
> > I cannot reproduce this with cygwin git head of cygwin-3_3-branch.
> > Is this MSYS2 specific problem?
> 
> You're right. I can't reproduce either. Only in my MSYS branch.
> 
> Can you give me some guidance how to debug it?

If you could identify which commit causes the issue, It would help.

-- 
Takashi Yano <takashi.yano@nifty.ne.jp>

Re: mintty crashes on Windows 7

[Orgad Shaneh]

On Tue, May 3, 2022 at 7:10 PM Takashi Yano <takashi.yano@nifty.ne.jp> wrote:
>
> On Tue, 3 May 2022 18:52:28 +0300
> Orgad Shaneh wrote:
> > On Tue, May 3, 2022 at 6:23 PM Takashi Yano <takashi.yano@nifty.ne.jp> wrote:
> > >
> > > On Tue, 3 May 2022 14:47:17 +0300
> > > Orgad Shaneh wrote:
> > > > Hi,
> > > >
> > > > Running `mintty ./bash` crashes on Windows 7 on cygwin-3_3-branch.
> > > >
> > > > Tested in MSYS2 on merge-3.3 branch from
> > > > https://github.com/orgads/msys2-runtime-1. It includes the tip of
> > > > cygwin-3_3-branch as of today (05827d2df8).
> > > >
> > > I cannot reproduce this with cygwin git head of cygwin-3_3-branch.
> > > Is this MSYS2 specific problem?
> >
> > You're right. I can't reproduce either. Only in my MSYS branch.
> >
> > Can you give me some guidance how to debug it?
>
> If you could identify which commit causes the issue, It would help.

0e1847fb87f5306cda6c022540560c5926627ae1 is the first bad commit
commit 0e1847fb87f5306cda6c022540560c5926627ae1
Author: Takashi Yano <takashi.yano@nifty.ne.jp>
Date:   Mon Feb 28 20:25:09 2022 +0900

    Cygwin: pty: Isolate CTRL_C_EVENTs between ptys.

    - With this patch, unique invisible consoles are created for each pty
      to isolate CTRL_C_EVENTs between ptys. This is necessary by Ctrl-C
      handling in fhandler_termios::process_sigs() for non-cygwin apps
      started in pty if the pseudo console is disabled.

 winsup/cygwin/fhandler_termios.cc |  6 ++----
 winsup/cygwin/fhandler_tty.cc     | 17 +++++++++++++++++
 winsup/cygwin/tty.cc              |  2 ++
 3 files changed, 21 insertions(+), 4 deletions(-)

I tried reverting this commit, and it fixed the crash indeed.

- Orgad

Re: mintty crashes on Windows 7

[Orgad Shaneh]

On Wed, May 4, 2022 at 12:33 PM Orgad Shaneh <orgads@gmail.com> wrote:
>
> On Tue, May 3, 2022 at 7:10 PM Takashi Yano <takashi.yano@nifty.ne.jp> wrote:
> >
> > On Tue, 3 May 2022 18:52:28 +0300
> > Orgad Shaneh wrote:
> > > On Tue, May 3, 2022 at 6:23 PM Takashi Yano <takashi.yano@nifty.ne.jp> wrote:
> > > >
> > > > On Tue, 3 May 2022 14:47:17 +0300
> > > > Orgad Shaneh wrote:
> > > > > Hi,
> > > > >
> > > > > Running `mintty ./bash` crashes on Windows 7 on cygwin-3_3-branch.
> > > > >
> > > > > Tested in MSYS2 on merge-3.3 branch from
> > > > > https://github.com/orgads/msys2-runtime-1. It includes the tip of
> > > > > cygwin-3_3-branch as of today (05827d2df8).
> > > > >
> > > > I cannot reproduce this with cygwin git head of cygwin-3_3-branch.
> > > > Is this MSYS2 specific problem?
> > >
> > > You're right. I can't reproduce either. Only in my MSYS branch.
> > >
> > > Can you give me some guidance how to debug it?
> >
> > If you could identify which commit causes the issue, It would help.
>
> 0e1847fb87f5306cda6c022540560c5926627ae1 is the first bad commit
> commit 0e1847fb87f5306cda6c022540560c5926627ae1
> Author: Takashi Yano <takashi.yano@nifty.ne.jp>
> Date:   Mon Feb 28 20:25:09 2022 +0900
>
>     Cygwin: pty: Isolate CTRL_C_EVENTs between ptys.
>
>     - With this patch, unique invisible consoles are created for each pty
>       to isolate CTRL_C_EVENTs between ptys. This is necessary by Ctrl-C
>       handling in fhandler_termios::process_sigs() for non-cygwin apps
>       started in pty if the pseudo console is disabled.
>
>  winsup/cygwin/fhandler_termios.cc |  6 ++----
>  winsup/cygwin/fhandler_tty.cc     | 17 +++++++++++++++++
>  winsup/cygwin/tty.cc              |  2 ++
>  3 files changed, 21 insertions(+), 4 deletions(-)
>
> I tried reverting this commit, and it fixed the crash indeed.
>
> - Orgad

Reduced the revert to this:
@@ -979,16 +979,10 @@ fhandler_pty_slave::open (int flags, mode_t)
        code does not work as expected because it calls Win32
        API directly rather than cygwin read()/write(). Due to
        this behaviour, protection based on attach_mutex does
        not take effect. */
     get_ttyp ()->need_invisible_console = true;
-  else if (_major (myself->ctty) != DEV_CONS_MAJOR
-    && (!get_ttyp ()->invisible_console_pid
-        || !pinfo (get_ttyp ()->invisible_console_pid)))
-    /* Create a new invisible console for each pty to isolate
-       CTRL_C_EVENTs between ptys. */
-    get_ttyp ()->need_invisible_console = true;
   else
     {
       acquire_attach_mutex (mutex_timeout);
       fhandler_console::need_invisible ();
       release_attach_mutex ();

- Orgad

Re: mintty crashes on Windows 7

[Takashi Yano]

On Wed, 4 May 2022 12:46:28 +0300
Orgad Shaneh wrote:
> On Wed, May 4, 2022 at 12:33 PM Orgad Shaneh <orgads@gmail.com> wrote:
> >
> > On Tue, May 3, 2022 at 7:10 PM Takashi Yano <takashi.yano@nifty.ne.jp> wrote:
> > >
> > > On Tue, 3 May 2022 18:52:28 +0300
> > > Orgad Shaneh wrote:
> > > > On Tue, May 3, 2022 at 6:23 PM Takashi Yano <takashi.yano@nifty.ne.jp> wrote:
> > > > >
> > > > > On Tue, 3 May 2022 14:47:17 +0300
> > > > > Orgad Shaneh wrote:
> > > > > > Hi,
> > > > > >
> > > > > > Running `mintty ./bash` crashes on Windows 7 on cygwin-3_3-branch.
> > > > > >
> > > > > > Tested in MSYS2 on merge-3.3 branch from
> > > > > > https://github.com/orgads/msys2-runtime-1. It includes the tip of
> > > > > > cygwin-3_3-branch as of today (05827d2df8).
> > > > > >
> > > > > I cannot reproduce this with cygwin git head of cygwin-3_3-branch.
> > > > > Is this MSYS2 specific problem?
> > > >
> > > > You're right. I can't reproduce either. Only in my MSYS branch.
> > > >
> > > > Can you give me some guidance how to debug it?
> > >
> > > If you could identify which commit causes the issue, It would help.
> >
> > 0e1847fb87f5306cda6c022540560c5926627ae1 is the first bad commit
> > commit 0e1847fb87f5306cda6c022540560c5926627ae1
> > Author: Takashi Yano <takashi.yano@nifty.ne.jp>
> > Date:   Mon Feb 28 20:25:09 2022 +0900
> >
> >     Cygwin: pty: Isolate CTRL_C_EVENTs between ptys.
> >
> >     - With this patch, unique invisible consoles are created for each pty
> >       to isolate CTRL_C_EVENTs between ptys. This is necessary by Ctrl-C
> >       handling in fhandler_termios::process_sigs() for non-cygwin apps
> >       started in pty if the pseudo console is disabled.
> >
> >  winsup/cygwin/fhandler_termios.cc |  6 ++----
> >  winsup/cygwin/fhandler_tty.cc     | 17 +++++++++++++++++
> >  winsup/cygwin/tty.cc              |  2 ++
> >  3 files changed, 21 insertions(+), 4 deletions(-)
> >
> > I tried reverting this commit, and it fixed the crash indeed.
> >
> > - Orgad
> 
> Reduced the revert to this:
> @@ -979,16 +979,10 @@ fhandler_pty_slave::open (int flags, mode_t)
>         code does not work as expected because it calls Win32
>         API directly rather than cygwin read()/write(). Due to
>         this behaviour, protection based on attach_mutex does
>         not take effect. */
>      get_ttyp ()->need_invisible_console = true;
> -  else if (_major (myself->ctty) != DEV_CONS_MAJOR
> -    && (!get_ttyp ()->invisible_console_pid
> -        || !pinfo (get_ttyp ()->invisible_console_pid)))
> -    /* Create a new invisible console for each pty to isolate
> -       CTRL_C_EVENTs between ptys. */
> -    get_ttyp ()->need_invisible_console = true;
>    else
>      {
>        acquire_attach_mutex (mutex_timeout);
>        fhandler_console::need_invisible ();
>        release_attach_mutex ();

A few things about this.

1) bash exits with exit code 127 for 'mintty bash'
2) 'mintty bash' does not work, but 'mintty ash' and 'mintty tcsh' work.

-- 
Takashi Yano <takashi.yano@nifty.ne.jp>

Re: mintty crashes on Windows 7

[Orgad Shaneh]

On Wed, May 4, 2022 at 2:16 PM Takashi Yano <takashi.yano@nifty.ne.jp> wrote:
>
> > Reduced the revert to this:
> > @@ -979,16 +979,10 @@ fhandler_pty_slave::open (int flags, mode_t)
> >         code does not work as expected because it calls Win32
> >         API directly rather than cygwin read()/write(). Due to
> >         this behaviour, protection based on attach_mutex does
> >         not take effect. */
> >      get_ttyp ()->need_invisible_console = true;
> > -  else if (_major (myself->ctty) != DEV_CONS_MAJOR
> > -    && (!get_ttyp ()->invisible_console_pid
> > -        || !pinfo (get_ttyp ()->invisible_console_pid)))
> > -    /* Create a new invisible console for each pty to isolate
> > -       CTRL_C_EVENTs between ptys. */
> > -    get_ttyp ()->need_invisible_console = true;
> >    else
> >      {
> >        acquire_attach_mutex (mutex_timeout);
> >        fhandler_console::need_invisible ();
> >        release_attach_mutex ();
>
> A few things about this.
>
> 1) bash exits with exit code 127 for 'mintty bash'
> 2) 'mintty bash' does not work, but 'mintty ash' and 'mintty tcsh' work.

Right. mintty dash also works.

Notice that I did *not* set enable_pcon (not supported on Win7 anyway).

- Orgad

Re: mintty crashes on Windows 7

[Takashi Yano]

On Wed, 4 May 2022 16:27:43 +0300
Orgad Shaneh  wrote:
> On Wed, May 4, 2022 at 2:16 PM Takashi Yano <takashi.yano@nifty.ne.jp> wrote:
> >
> > > Reduced the revert to this:
> > > @@ -979,16 +979,10 @@ fhandler_pty_slave::open (int flags, mode_t)
> > >         code does not work as expected because it calls Win32
> > >         API directly rather than cygwin read()/write(). Due to
> > >         this behaviour, protection based on attach_mutex does
> > >         not take effect. */
> > >      get_ttyp ()->need_invisible_console = true;
> > > -  else if (_major (myself->ctty) != DEV_CONS_MAJOR
> > > -    && (!get_ttyp ()->invisible_console_pid
> > > -        || !pinfo (get_ttyp ()->invisible_console_pid)))
> > > -    /* Create a new invisible console for each pty to isolate
> > > -       CTRL_C_EVENTs between ptys. */
> > > -    get_ttyp ()->need_invisible_console = true;
> > >    else
> > >      {
> > >        acquire_attach_mutex (mutex_timeout);
> > >        fhandler_console::need_invisible ();
> > >        release_attach_mutex ();
> >
> > A few things about this.
> >
> > 1) bash exits with exit code 127 for 'mintty bash'
> > 2) 'mintty bash' does not work, but 'mintty ash' and 'mintty tcsh' work.
> 
> Right. mintty dash also works.
> 
> Notice that I did *not* set enable_pcon (not supported on Win7 anyway).

Even without that commit (and also with msys2 3.3.4 release),
something wrong with msys2 in Windows 7.

If you run script command (/usr/bin/script), bash crashes in
similar way. typescript file generated by script command is
as follows.

Script started on 2022-05-05 10:12:28+09:00 [TERM="cygwin" TTY="/dev/cons0" COLUMNS="80" LINES="25"]

Script done on 2022-05-05 10:12:29+09:00 [COMMAND_EXIT_CODE="127"]

bash also exited with exit code 127.

-- 
Takashi Yano <takashi.yano@nifty.ne.jp>

Re: mintty crashes on Windows 7

[Takashi Yano]

On Thu, 5 May 2022 10:20:45 +0900
Takashi Yano <takashi.yano@nifty.ne.jp> wrote:

> On Wed, 4 May 2022 16:27:43 +0300
> Orgad Shaneh  wrote:
> > On Wed, May 4, 2022 at 2:16 PM Takashi Yano <takashi.yano@nifty.ne.jp> wrote:
> > >
> > > > Reduced the revert to this:
> > > > @@ -979,16 +979,10 @@ fhandler_pty_slave::open (int flags, mode_t)
> > > >         code does not work as expected because it calls Win32
> > > >         API directly rather than cygwin read()/write(). Due to
> > > >         this behaviour, protection based on attach_mutex does
> > > >         not take effect. */
> > > >      get_ttyp ()->need_invisible_console = true;
> > > > -  else if (_major (myself->ctty) != DEV_CONS_MAJOR
> > > > -    && (!get_ttyp ()->invisible_console_pid
> > > > -        || !pinfo (get_ttyp ()->invisible_console_pid)))
> > > > -    /* Create a new invisible console for each pty to isolate
> > > > -       CTRL_C_EVENTs between ptys. */
> > > > -    get_ttyp ()->need_invisible_console = true;
> > > >    else
> > > >      {
> > > >        acquire_attach_mutex (mutex_timeout);
> > > >        fhandler_console::need_invisible ();
> > > >        release_attach_mutex ();
> > >
> > > A few things about this.
> > >
> > > 1) bash exits with exit code 127 for 'mintty bash'
> > > 2) 'mintty bash' does not work, but 'mintty ash' and 'mintty tcsh' work.
> > 
> > Right. mintty dash also works.
> > 
> > Notice that I did *not* set enable_pcon (not supported on Win7 anyway).
> 
> Even without that commit (and also with msys2 3.3.4 release),
> something wrong with msys2 in Windows 7.
> 
> If you run script command (/usr/bin/script), bash crashes in
> similar way. typescript file generated by script command is
> as follows.
> 
> Script started on 2022-05-05 10:12:28+09:00 [TERM="cygwin" TTY="/dev/cons0" COLUMNS="80" LINES="25"]
> 
> Script done on 2022-05-05 10:12:29+09:00 [COMMAND_EXIT_CODE="127"]
> 
> bash also exited with exit code 127.

Ah, this only occurs if script command is started in console.

-- 
Takashi Yano <takashi.yano@nifty.ne.jp>

Re: mintty crashes on Windows 7

[Takashi Yano]

On Thu, 5 May 2022 10:27:24 +0900
Takashi Yano wrote:
> On Thu, 5 May 2022 10:20:45 +0900
> Takashi Yano wrote:
> 
> > On Wed, 4 May 2022 16:27:43 +0300
> > Orgad Shaneh  wrote:
> > > On Wed, May 4, 2022 at 2:16 PM Takashi Yano <takashi.yano@nifty.ne.jp> wrote:
> > > >
> > > > > Reduced the revert to this:
> > > > > @@ -979,16 +979,10 @@ fhandler_pty_slave::open (int flags, mode_t)
> > > > >         code does not work as expected because it calls Win32
> > > > >         API directly rather than cygwin read()/write(). Due to
> > > > >         this behaviour, protection based on attach_mutex does
> > > > >         not take effect. */
> > > > >      get_ttyp ()->need_invisible_console = true;
> > > > > -  else if (_major (myself->ctty) != DEV_CONS_MAJOR
> > > > > -    && (!get_ttyp ()->invisible_console_pid
> > > > > -        || !pinfo (get_ttyp ()->invisible_console_pid)))
> > > > > -    /* Create a new invisible console for each pty to isolate
> > > > > -       CTRL_C_EVENTs between ptys. */
> > > > > -    get_ttyp ()->need_invisible_console = true;
> > > > >    else
> > > > >      {
> > > > >        acquire_attach_mutex (mutex_timeout);
> > > > >        fhandler_console::need_invisible ();
> > > > >        release_attach_mutex ();
> > > >
> > > > A few things about this.
> > > >
> > > > 1) bash exits with exit code 127 for 'mintty bash'
> > > > 2) 'mintty bash' does not work, but 'mintty ash' and 'mintty tcsh' work.
> > > 
> > > Right. mintty dash also works.
> > > 
> > > Notice that I did *not* set enable_pcon (not supported on Win7 anyway).
> > 
> > Even without that commit (and also with msys2 3.3.4 release),
> > something wrong with msys2 in Windows 7.
> > 
> > If you run script command (/usr/bin/script), bash crashes in
> > similar way. typescript file generated by script command is
> > as follows.
> > 
> > Script started on 2022-05-05 10:12:28+09:00 [TERM="cygwin" TTY="/dev/cons0" COLUMNS="80" LINES="25"]
> > 
> > Script done on 2022-05-05 10:12:29+09:00 [COMMAND_EXIT_CODE="127"]
> > 
> > bash also exited with exit code 127.
> 
> Ah, this only occurs if script command is started in console.

And also this is caused by:
get_ttyp ()->need_invisible_console = true;
at another place.

-- 
Takashi Yano <takashi.yano@nifty.ne.jp>

Re: mintty crashes on Windows 7

[Takashi Yano]

On Thu, 5 May 2022 10:59:54 +0900
Takashi Yano wrote:
> On Thu, 5 May 2022 10:27:24 +0900
> Takashi Yano wrote:
> > On Thu, 5 May 2022 10:20:45 +0900
> > Takashi Yano wrote:
> > 
> > > On Wed, 4 May 2022 16:27:43 +0300
> > > Orgad Shaneh  wrote:
> > > > On Wed, May 4, 2022 at 2:16 PM Takashi Yano <takashi.yano@nifty.ne.jp> wrote:
> > > > >
> > > > > > Reduced the revert to this:
> > > > > > @@ -979,16 +979,10 @@ fhandler_pty_slave::open (int flags, mode_t)
> > > > > >         code does not work as expected because it calls Win32
> > > > > >         API directly rather than cygwin read()/write(). Due to
> > > > > >         this behaviour, protection based on attach_mutex does
> > > > > >         not take effect. */
> > > > > >      get_ttyp ()->need_invisible_console = true;
> > > > > > -  else if (_major (myself->ctty) != DEV_CONS_MAJOR
> > > > > > -    && (!get_ttyp ()->invisible_console_pid
> > > > > > -        || !pinfo (get_ttyp ()->invisible_console_pid)))
> > > > > > -    /* Create a new invisible console for each pty to isolate
> > > > > > -       CTRL_C_EVENTs between ptys. */
> > > > > > -    get_ttyp ()->need_invisible_console = true;
> > > > > >    else
> > > > > >      {
> > > > > >        acquire_attach_mutex (mutex_timeout);
> > > > > >        fhandler_console::need_invisible ();
> > > > > >        release_attach_mutex ();
> > > > >
> > > > > A few things about this.
> > > > >
> > > > > 1) bash exits with exit code 127 for 'mintty bash'
> > > > > 2) 'mintty bash' does not work, but 'mintty ash' and 'mintty tcsh' work.
> > > > 
> > > > Right. mintty dash also works.
> > > > 
> > > > Notice that I did *not* set enable_pcon (not supported on Win7 anyway).
> > > 
> > > Even without that commit (and also with msys2 3.3.4 release),
> > > something wrong with msys2 in Windows 7.
> > > 
> > > If you run script command (/usr/bin/script), bash crashes in
> > > similar way. typescript file generated by script command is
> > > as follows.
> > > 
> > > Script started on 2022-05-05 10:12:28+09:00 [TERM="cygwin" TTY="/dev/cons0" COLUMNS="80" LINES="25"]
> > > 
> > > Script done on 2022-05-05 10:12:29+09:00 [COMMAND_EXIT_CODE="127"]
> > > 
> > > bash also exited with exit code 127.
> > 
> > Ah, this only occurs if script command is started in console.
> 
> And also this is caused by:
> get_ttyp ()->need_invisible_console = true;
> at another place.

I found that bash crashes at:
      h = GetProcessWindowStation ();
in fhandler_console::need_invisible().

-- 
Takashi Yano <takashi.yano@nifty.ne.jp>

Re: mintty crashes on Windows 7

[Takashi Yano]

On Thu, 5 May 2022 12:33:07 +0900
Takashi Yano wrote:
> On Thu, 5 May 2022 10:59:54 +0900
> Takashi Yano wrote:
> > On Thu, 5 May 2022 10:27:24 +0900
> > Takashi Yano wrote:
> > > On Thu, 5 May 2022 10:20:45 +0900
> > > Takashi Yano wrote:
> > > 
> > > > On Wed, 4 May 2022 16:27:43 +0300
> > > > Orgad Shaneh  wrote:
> > > > > On Wed, May 4, 2022 at 2:16 PM Takashi Yano <takashi.yano@nifty.ne.jp> wrote:
> > > > > >
> > > > > > > Reduced the revert to this:
> > > > > > > @@ -979,16 +979,10 @@ fhandler_pty_slave::open (int flags, mode_t)
> > > > > > >         code does not work as expected because it calls Win32
> > > > > > >         API directly rather than cygwin read()/write(). Due to
> > > > > > >         this behaviour, protection based on attach_mutex does
> > > > > > >         not take effect. */
> > > > > > >      get_ttyp ()->need_invisible_console = true;
> > > > > > > -  else if (_major (myself->ctty) != DEV_CONS_MAJOR
> > > > > > > -    && (!get_ttyp ()->invisible_console_pid
> > > > > > > -        || !pinfo (get_ttyp ()->invisible_console_pid)))
> > > > > > > -    /* Create a new invisible console for each pty to isolate
> > > > > > > -       CTRL_C_EVENTs between ptys. */
> > > > > > > -    get_ttyp ()->need_invisible_console = true;
> > > > > > >    else
> > > > > > >      {
> > > > > > >        acquire_attach_mutex (mutex_timeout);
> > > > > > >        fhandler_console::need_invisible ();
> > > > > > >        release_attach_mutex ();
> > > > > >
> > > > > > A few things about this.
> > > > > >
> > > > > > 1) bash exits with exit code 127 for 'mintty bash'
> > > > > > 2) 'mintty bash' does not work, but 'mintty ash' and 'mintty tcsh' work.
> > > > > 
> > > > > Right. mintty dash also works.
> > > > > 
> > > > > Notice that I did *not* set enable_pcon (not supported on Win7 anyway).
> > > > 
> > > > Even without that commit (and also with msys2 3.3.4 release),
> > > > something wrong with msys2 in Windows 7.
> > > > 
> > > > If you run script command (/usr/bin/script), bash crashes in
> > > > similar way. typescript file generated by script command is
> > > > as follows.
> > > > 
> > > > Script started on 2022-05-05 10:12:28+09:00 [TERM="cygwin" TTY="/dev/cons0" COLUMNS="80" LINES="25"]
> > > > 
> > > > Script done on 2022-05-05 10:12:29+09:00 [COMMAND_EXIT_CODE="127"]
> > > > 
> > > > bash also exited with exit code 127.
> > > 
> > > Ah, this only occurs if script command is started in console.
> > 
> > And also this is caused by:
> > get_ttyp ()->need_invisible_console = true;
> > at another place.
> 
> I found that bash crashes at:
>       h = GetProcessWindowStation ();
> in fhandler_console::need_invisible().

I downloaded bash source files from:
https://git.savannah.gnu.org/git/bash.git
and built it by:
./configure && make
and replaced /usr/bin/bash.

Then the issue disappeared.
Now, I start to suspect the issue is a bug of msys2 bash package.

-- 
Takashi Yano <takashi.yano@nifty.ne.jp>

Re: mintty crashes on Windows 7

[Takashi Yano]

On Thu, 5 May 2022 13:41:20 +0900
Takashi Yano wrote:
> I downloaded bash source files from:
> https://git.savannah.gnu.org/git/bash.git
> and built it by:
> ./configure && make
> and replaced /usr/bin/bash.
> 
> Then the issue disappeared.
> Now, I start to suspect the issue is a bug of msys2 bash package.

I tried several shells and results are as follows.

tcsh: ok
ash: ok
dash: ok
zsh: ok
ksh: ok
fish: ok
bash (build from original source): ok

bash (from msys2 package): ng

Only bash in msys2 package fails.

-- 
Takashi Yano <takashi.yano@nifty.ne.jp>

Re: mintty crashes on Windows 7

[Takashi Yano]

On Thu, 5 May 2022 15:44:40 +0900
Takashi Yano wrote:
> On Thu, 5 May 2022 13:41:20 +0900
> Takashi Yano wrote:
> > I downloaded bash source files from:
> > https://git.savannah.gnu.org/git/bash.git
> > and built it by:
> > ./configure && make
> > and replaced /usr/bin/bash.
> > 
> > Then the issue disappeared.
> > Now, I start to suspect the issue is a bug of msys2 bash package.
> 
> I tried several shells and results are as follows.
> 
> tcsh: ok
> ash: ok
> dash: ok
> zsh: ok
> ksh: ok
> fish: ok
> bash (build from original source): ok
> 
> bash (from msys2 package): ng
> 
> Only bash in msys2 package fails.

I identified the difference which causes the issue
between bash built from original source and msys2 bash.

If --enable-readline and --with-installed-readline is specified
to configure, the problem causes even with bash built from original
source.

Also, removing --with-installed-readline from configure and removing
READLINE_LDFLAGS= from make in PKGBUILD of MSYS2 bash package solves
the issue.

So it seems to be a readline problem, not a bug in bash itself.

-- 
Takashi Yano <takashi.yano@nifty.ne.jp>

Re: mintty crashes on Windows 7

[Orgad Shaneh]

On Fri, May 6, 2022 at 1:49 AM Takashi Yano <takashi.yano@nifty.ne.jp> wrote:
> > Only bash in msys2 package fails.
>
> I identified the difference which causes the issue
> between bash built from original source and msys2 bash.
>
> If --enable-readline and --with-installed-readline is specified
> to configure, the problem causes even with bash built from original
> source.
>
> Also, removing --with-installed-readline from configure and removing
> READLINE_LDFLAGS= from make in PKGBUILD of MSYS2 bash package solves
> the issue.
>
> So it seems to be a readline problem, not a bug in bash itself.

Adding @Johannes Schindelin to the loop.

Thanks for your investigation so far.

So how do we proceed, assuming MSYS project does want to use the
installed readline? I checked the readline PKGBUILD, and it only has
very basic patches, none of them looks suspicious.

Is (external) readline not used in cygwin/bash?

Do you suggest that there's a bug in readline, that your change in the
runtime uncovered? Or is it the other way around?

What might we break if we revert the part I referenced earlier in
fhandler_tty.cc?

- Orgad

Re: mintty crashes on Windows 7

[Brian Inglis]

On 2022-05-06 08:05, Orgad Shaneh wrote:
> On Fri, May 6, 2022 at 1:49 AM Takashi Yano wrote:
>>> Only bash in msys2 package fails.

>> I identified the difference which causes the issue
>> between bash built from original source and msys2 bash.
>> If --enable-readline and --with-installed-readline is specified
>> to configure, the problem causes even with bash built from original
>> source.
>> Also, removing --with-installed-readline from configure and removing
>> READLINE_LDFLAGS= from make in PKGBUILD of MSYS2 bash package solves
>> the issue.
>> So it seems to be a readline problem, not a bug in bash itself.

> Adding @Johannes Schindelin to the loop.
> Thanks for your investigation so far.
> So how do we proceed, assuming MSYS project does want to use the
> installed readline? I checked the readline PKGBUILD, and it only has
> very basic patches, none of them looks suspicious.
> Is (external) readline not used in cygwin/bash?
> Do you suggest that there's a bug in readline, that your change in the
> runtime uncovered? Or is it the other way around?
> What might we break if we revert the part I referenced earlier in
> fhandler_tty.cc?

Cygwin bash and other packages use the latest readline 8.1 library package:

	$ cygstart https://cygwin.com/packages/summary/readline-src.html

	$ cygcheck -l libreadline7
	/usr/bin/cygreadline7.dll
	/usr/bin/cyghistory7.dll

What readline does MSYS2 bash use - it requires libreadline-devel to 
build, but does not have any readline dependency, nor is libreadline 
required by bash:

	https://packages.msys2.org/package/bash
	https://packages.msys2.org/package/libreadline


-- 
Take care. Thanks, Brian Inglis, Calgary, Alberta, Canada

This email may be disturbing to some readers as it contains
too much technical detail. Reader discretion is advised.
[Data in binary units and prefixes, physical quantities in SI.]

Re: mintty crashes on Windows 7

[Takashi Yano]

On Fri, 6 May 2022 17:05:02 +0300
Orgad Shaneh wrote:
> On Fri, May 6, 2022 at 1:49 AM Takashi Yano <takashi.yano@nifty.ne.jp> wrote:
> > > Only bash in msys2 package fails.
> >
> > I identified the difference which causes the issue
> > between bash built from original source and msys2 bash.
> >
> > If --enable-readline and --with-installed-readline is specified
> > to configure, the problem causes even with bash built from original
> > source.
> >
> > Also, removing --with-installed-readline from configure and removing
> > READLINE_LDFLAGS= from make in PKGBUILD of MSYS2 bash package solves
> > the issue.
> >
> > So it seems to be a readline problem, not a bug in bash itself.
> 
> Adding @Johannes Schindelin to the loop.
> 
> Thanks for your investigation so far.
> 
> So how do we proceed, assuming MSYS project does want to use the
> installed readline? I checked the readline PKGBUILD, and it only has
> very basic patches, none of them looks suspicious.
> 
> Is (external) readline not used in cygwin/bash?

In cygwin the issue does not occur even with bash
configured with --enable-readline and --with-installed-readline.

> Do you suggest that there's a bug in readline, that your change in the
> runtime uncovered? Or is it the other way around?

I suspect it's the former.

> What might we break if we revert the part I referenced earlier in
> fhandler_tty.cc?

If you revert that commit, non-cygwin app recieves ctrl-c twice
in the following scenario.

1) Start mintty
2) Run cmd.exe
3) Run script command
4) Run non-cygwin app
5) Press ctrl-c

This is becase ctrl-c is processed by both mintty pty master
and script pty master.

-- 
Takashi Yano <takashi.yano@nifty.ne.jp>

Re: mintty crashes on Windows 7

[Johannes Schindelin]

Hi Orgad,

On Fri, 6 May 2022, Orgad Shaneh wrote:

> Adding @Johannes Schindelin to the loop.

Thank you, but that unfortunately does not work on this list. Due to the
policy to never reply-to-all, the subsequent replies immediately lost me.

The reason why MSYS2's Bash does not depend on the `libreadline` package
is that `bash.exe` is linked _statically_ to that library.

Takashi, for the record, I find it hard to believe that the bug is
libreadline's because Orgad's scenario works if he reverts that patch in
the _MSYS2 runtime_, _and_ it is rather dubious that libreadline would
potentially do anything that makes a call to `GetProcessWindowStation()`
not fail but _crash_.

These console patches sure are gifts that keep on giving.

Ciao,
Dscho

Re: mintty crashes on Windows 7

[Takashi Yano]

On Fri, 6 May 2022 21:16:10 +0200 (CEST)
Johannes Schindelin wrote:
> Takashi, for the record, I find it hard to believe that the bug is
> libreadline's because Orgad's scenario works if he reverts that patch in
> the _MSYS2 runtime_, _and_ it is rather dubious that libreadline would
> potentially do anything that makes a call to `GetProcessWindowStation()`
> not fail but _crash_.

I found the following test case also crashes with that commit.

1) Compile rl_stc.c with gcc rl_stc.c -lreadline -o rl_stc.c
2) mintty --hold always ./rl_stc

/* rl_stc.c */
#include <stdio.h>
#include <stdlib.h>
#include <readline/readline.h>

int main(int argc, char *argv[])
{
    char *str;
    if (argc > 1) {
        str = readline(">> ");
        printf("%s\n", str);
        free(str);
    }
    return 0;
}

In this test case, no args is specified, so readline() is
not called. However, this crashes indeed. This means just
loading msys-readline8.dll causes the crash.

-- 
Takashi Yano <takashi.yano@nifty.ne.jp>

Re: mintty crashes on Windows 7

[Takashi Yano]

On Sat, 7 May 2022 05:13:23 +0900
Takashi Yano wrote:
> On Fri, 6 May 2022 21:16:10 +0200 (CEST)
> Johannes Schindelin wrote:
> > Takashi, for the record, I find it hard to believe that the bug is
> > libreadline's because Orgad's scenario works if he reverts that patch in
> > the _MSYS2 runtime_, _and_ it is rather dubious that libreadline would
> > potentially do anything that makes a call to `GetProcessWindowStation()`
> > not fail but _crash_.
> 
> I found the following test case also crashes with that commit.
> 
> 1) Compile rl_stc.c with gcc rl_stc.c -lreadline -o rl_stc.c
> 2) mintty --hold always ./rl_stc
> 
> /* rl_stc.c */
> #include <stdio.h>
> #include <stdlib.h>
> #include <readline/readline.h>
> 
> int main(int argc, char *argv[])
> {
>     char *str;
>     if (argc > 1) {
>         str = readline(">> ");
>         printf("%s\n", str);
>         free(str);
>     }
>     return 0;
> }
> 
> In this test case, no args is specified, so readline() is
> not called. However, this crashes indeed. This means just
> loading msys-readline8.dll causes the crash.

1') gcc --static rl_stc.c -lreadline -lncurses -o rl_stc.c
2) mintty --hold always ./rl_stc

also causes crash. In this case, no code from libreadline is
executed, I think. Why this triggers GetProcessWindowStation()
crash?

-- 
Takashi Yano <takashi.yano@nifty.ne.jp>

Re: mintty crashes on Windows 7

[Takashi Yano]

On Sat, 7 May 2022 09:20:51 +0900
Takashi Yano wrote:
> On Sat, 7 May 2022 05:13:23 +0900
> Takashi Yano wrote:
> > On Fri, 6 May 2022 21:16:10 +0200 (CEST)
> > Johannes Schindelin wrote:
> > > Takashi, for the record, I find it hard to believe that the bug is
> > > libreadline's because Orgad's scenario works if he reverts that patch in
> > > the _MSYS2 runtime_, _and_ it is rather dubious that libreadline would
> > > potentially do anything that makes a call to `GetProcessWindowStation()`
> > > not fail but _crash_.
> > 
> > I found the following test case also crashes with that commit.
> > 
> > 1) Compile rl_stc.c with gcc rl_stc.c -lreadline -o rl_stc.c
> > 2) mintty --hold always ./rl_stc
> > 
> > /* rl_stc.c */
> > #include <stdio.h>
> > #include <stdlib.h>
> > #include <readline/readline.h>
> > 
> > int main(int argc, char *argv[])
> > {
> >     char *str;
> >     if (argc > 1) {
> >         str = readline(">> ");
> >         printf("%s\n", str);
> >         free(str);
> >     }
> >     return 0;
> > }
> > 
> > In this test case, no args is specified, so readline() is
> > not called. However, this crashes indeed. This means just
> > loading msys-readline8.dll causes the crash.
> 
> 1') gcc --static rl_stc.c -lreadline -lncurses -o rl_stc.c
> 2) mintty --hold always ./rl_stc
> 
> also causes crash. In this case, no code from libreadline is
> executed, I think. Why this triggers GetProcessWindowStation()
> crash?

I have pushed two patches to cygwin-3_3-branch. I am not sure
why, but the issue (bash with readline crash) seems to disappear.

Could you please try?

-- 
Takashi Yano <takashi.yano@nifty.ne.jp>

Re: mintty crashes on Windows 7

[Orgad Shaneh]

On Sun, May 8, 2022 at 2:09 PM Takashi Yano <takashi.yano@nifty.ne.jp> wrote:
> I have pushed two patches to cygwin-3_3-branch. I am not sure
> why, but the issue (bash with readline crash) seems to disappear.
>
> Could you please try?

It no longer crashes with these fixes.

Thanks!

- Orgad