RE: Git 2.54.0-rc1, subtests of t5310, t5326, t5327

[<rsbecker@nexbridge.com>]

On April 8, 2026 10:54 AM, I wrote:
>To: 'Jeff King' <peff@peff.net>
>Cc: git@vger.kernel.org
>Subject: RE: Git 2.54.0-rc1, subtests of t5310, t5326, t5327
>
>On April 8, 2026 12:17 AM, Jeff King wrote:
>>On Tue, Apr 07, 2026 at 07:29:50PM -0400, rsbecker@nexbridge.com wrote:
>>
>>> I can getting numerous issues in t5310, t5326, t2527 relating to the
>>> following use of --git-dir:
>>>
>>> In t5310:
>>> fatal: not a git repository: 'clone.git'
>>> not ok 55 - fetch (full bitmap)
>>> #
>>> #                       git --git-dir=clone.git fetch origin second:second
>>> &&
>>> #                       git rev-parse HEAD >expect &&
>>> #                       git --git-dir=clone.git rev-parse HEAD >actual &&
>>> #                       test_cmp expect actual
>>> #
>>
>>This test hasn't changed recently. The clone.git directory should have been created
>>by an earlier test. Can you try running with "-i" and make sure that this is the first
>>failing test, and we didn't fail earlier?
>>
>>Especially because...
>>
>>> In t5326 and t5327:
>>> fatal: writev error: Invalid function argument
>>> fetch-pack: unexpected disconnect while reading sideband packet
>>> fatal: early EOF
>>> fatal: fetch-pack: invalid index-pack output not ok 24 - clone from
>>> bitmapped repository #
>>> #                       rm -fr clone.git &&
>>> #                       git clone --no-local --bare . clone.git &&
>>> #                       git rev-parse HEAD >expect &&
>>> #                       git --git-dir=clone.git rev-parse HEAD >actual &&
>>> #                       test_cmp expect actual
>>> #
>>
>>...it looks less like --git-dir is a problem here, and more like the introduction of
>>writev() is. It is now used in sideband_send(), so it seems plausible that a similar
>>failure might have broken the git-clone operation that the other test was using to
>>create clone.git.
>>
>>As for why writev() is failing, I don't know. If it were totally broken on your system
>>I'd expect almost everything to be failing. But maybe try building with "make
>>NO_WRITEV=Nope" and see if that makes the problems go away? The compat
>>implementation just does a series of write() calls, which is what send_sideband()
>>was doing before.
>
>First fail is as follows in subtest 25:
>
>expecting success of 5310.25 'clone from bitmapped repository':
>                rm -fr clone.git &&
>                git clone --no-local --bare . clone.git &&
>                git rev-parse HEAD >expect &&
>                git --git-dir=clone.git rev-parse HEAD >actual &&
>                test_cmp expect actual
>
>Cloning into bare repository 'clone.git'...
>remote: Enumerating objects: 629, done.
>fatal: writev error: Invalid function argument
>fetch-pack: unexpected disconnect while reading sideband packet
>fatal: early EOF
>fatal: fetch-pack: invalid index-pack output
>not ok 25 - clone from bitmapped repository
>
>I think the invalid function argument maybe an ioctl or socketioctl not supported for
>the file type.

This is also impacting t5608 and t7700. Anywhere where writev() is used, seemingly. We went through
MAX_IO_SIZE issues years ago, instead of using ssize_t as a basis of how big communication is. I think
writev() is not valid. It worked on Lunix, but had issues elsewhere. This broke the compat layer.