sh Stack Overflow when pushing subtree. (2.38.1 works 2.39.1 does not)

[josh08287]

• [x ] I was not able to find an open or closed issue matching what I'm seeing

Setup

• Which version of Git for Windows are you using? Is it 32-bit or 64-bit?

$ git --version --build-options

**
WORKING VERSION 
$ git --version --build-options
git version 2.38.1.windows.1
cpu: x86_64
built from commit: b85c8f604d375d4d773a36842964e8a7ec056aae
sizeof-long: 4
sizeof-size_t: 8
shell-path: /bin/sh
feature: fsmonitor--daemon 

NOT WORKING VERSION
$ git --version --build-options
git version 2.39.1.windows.1
cpu: x86_64
built from commit: b03dafd9c26b06c92d509a07ab01b01e6d0d85ee
sizeof-long: 4
sizeof-size_t: 8
shell-path: /bin/sh
feature: fsmonitor--daemon

**

• Which version of Windows are you running? Vista, 7, 8, 10? Is it 32-bit or 64-bit?

$ cmd.exe /c ver

** Microsoft Windows [Version 10.0.19045.2486] **

• What options did you set as part of the installation? Or did you choose the
  defaults?

# One of the following:
> type "C:\Program Files\Git\etc\install-options.txt"
> type "C:\Program Files (x86)\Git\etc\install-options.txt"
> type "%USERPROFILE%\AppData\Local\Programs\Git\etc\install-options.txt"
> type "$env:USERPROFILE\AppData\Local\Programs\Git\etc\install-options.txt"
$ cat /etc/install-options.txt

** Editor Option: VIM
Custom Editor Path:
Default Branch Option:
Path Option: Cmd
SSH Option: OpenSSH
Tortoise Option: false
CURL Option: WinSSL
CRLF Option: CRLFAlways
Bash Terminal Option: MinTTY
Git Pull Behavior Option: Merge
Use Credential Manager: Enabled
Performance Tweaks FSCache: Enabled
Enable Symlinks: Enabled
Enable Pseudo Console Support: Disabled
Enable FSMonitor: Disabled **

• Any other interesting things about your environment that might be related
  to the issue you're seeing?

** we're using subtrees**

Details

• Which terminal/shell are you running Git from? e.g Bash/CMD/PowerShell/other

** mingw bash **

• What commands did you run to trigger this issue? If you can provide a
  Minimal, Complete, and Verifiable example
  this will help us understand the issue.

** git subtree push --prefix=<prefix> <repo> <branch> **

• What did you expect to occur after running these commands?

** subtree split and push happens successfully **

• What actually happened instead?

** sh process crashes with stackdump of:

Exception: STATUS_STACK_OVERFLOW at rip=7FFFA5EDC2B3
rax=0000CCEBCBC31338 rbx=00000000FFE054A0 rcx=0000000000000340
rdx=0000000000000000 rsi=00000000FFE04920 rdi=0000000000000000
r8 =000000021023EEA0 r9 =00000000FFE047B8 r10=0000000000000000
r11=0000000000000246 r12=00000000FFE04910 r13=0000000000000000
r14=000000021023EEA0 r15=0000000000000000
rbp=00000000FFE03E00 rsp=00000000FFE03D00
program=C:\Program Files\Git\usr\bin\sh.exe, pid 1103, thread 
cs=0033 ds=002B es=002B fs=0053 gs=002B ss=002B
Stack trace:
Frame        Function    Args
000FFE03E00  7FFFA5EDC2B3 (000FFE054A0, 00000000000, 00000000000, 00000000000)
000FFE04560  7FFFA5EDB496 (1D90EF56088401F, 00000000000, 00000000001, 000FFE048A8)
000FFE04950  7FFFA5EDBD87 (00000000008, 00000000046, 7FFFA72EB238, 00000000000)
000FFE054C0  7FFFA5892993 (00000000420, 00000000420, 000007A1670, 000FFE05E90)
00000000420  7FFFA4CA9215 (000FFE06D90, 000000002B8, 002100D0000, 00800018190)
00000000420  7FFFA4CA73C6 (002100577DE, 0021023EEA0, 000FFE06F58, 00000000001)
00000000420  7FFFA587D374 (000FFE07270, 000FFE07280, 00000000000, 00000000001)
00000000420  002100D3A1F (000FFE06FB0, 00000000000, 00000000000, 00000000000)
001004F84B7  002100D4638 (008000E8E50, 008000E8E50, 00000000008, 000FFE07270)
001004F84B7  002100D496A (000FFE07328, 000FFE07514, 00000000008, 000000E8E50)
001004F84B7  00210199A7B (000FFE07328, 000FFE07514, 00000000008, 000000E8E50)
001004F84B7  00000BFE458 (000FFE07514, 00000000008, 000000E8E50, 008000E8E50)
001004F84B7  00800069700 (00000000008, 000000E8E50, 008000E8E50, 00000084002)
001004F84B7  000FFE07328 (000000E8E50, 008000E8E50, 00000084002, 00000000000)
001004F84B7  000FFE07514 (008000E8E50, 00000084002, 00000000000, 00000004000)
001004F84B7  00000000008 (00000084002, 00000000000, 00000004000, 00100479620)
001004F84B7  000000E8E50 (00000000000, 00000004000, 00100479620, 00000000001)
001004F84B7  008000E8E50 (00000004000, 00100479620, 00000000001, 00000000008)
001004F84B7  00000084002 (00000004000, 00100479620, 00000000001, 00000000008)
End of stack trace

**

• If the problem was occurring with a specific repository, can you provide the
  URL to that repository to help us with testing?

** private repos**

It actually looks like it's the bundled sh.exe that is having a stack overflow due to the subtree split script. Is it possible to build the sh.exe with a larger stack size or fix the subtree script such that it isn't as recursvie? Our repo is only ~3000 commits with about 188 of those being split into the subtree repo.

[josh08287]

#2529 seems to be the same issue. Note this was also an issue with 2.28.x .

[dscho]

I believe that gitgitgadget#493 might be the clue to fixing this. @josh08287 if you have any resources to try this out (e.g. by rebasing those patches on top of Git for Windows v2.39.1 and running the git-artifacts workflow manually to generate an installer/Portable Git to test), it might help to move this stalled effort forward.

[josh08287]

@dscho - I'm not sure that this is the exact problem I'm having, however I'll give it a shot. I'm working out of a brand new cloned repo each time I try to push to a subtree, so there is no subtree cache to be dealt with.
Since I'm not set up at all to run any workflows here I'm going to just apply the patch to the script manually on my latest installed version.

[josh08287]

Applying the git-subtree.sh from the referenced PR directly to my git-subtree script in my installed version of 2.39.1 and it successfully completes the subtree push without crashing.
I still get a bad address error shown below, but I've always gotten this and it has worked:
C:/Program Files/Git/mingw64/libexec/git-core\git-subtree: line 643: /mingw64/libexec/git-core/git: Bad address

Let me know if there's anything else I can do to help speed this stuff along.

[dscho]

Let me know if there's anything else I can do to help speed this stuff along.

I guess you could prod @tqc to see whether they can pick up from where I left off. Or you could give it a try yourself to split this commit and add a regression test...