Re: [PATCH] Use DatumGetUInt32() for dsm_attach() in test_shm_mq_main()

Nathan Bossart <nathandbossart@gmail.com>

From: Nathan Bossart <nathandbossart@gmail.com>
To: Jianghua Yang <yjhjstz@gmail.com>
Cc: pgsql-hackers@lists.postgresql.org
Date: 2025-06-26T20:30:57Z
Lists: pgsql-hackers
On Thu, Jun 26, 2025 at 12:51:10PM -0700, Jianghua Yang wrote:
> The argument passed to `dsm_attach()` is expected to be a `uint32`, but the
> code currently uses `DatumGetInt32()` to extract it from the `Datum`
> argument. This can lead to incorrect behavior when the high bit is set, as
> 'unable to map dynamic shared memory segment'.

I'm not sure this actually causes any problems in practice because
dsm_attach() treats its argument as unsigned.  In any case, I've never seen
this test fail like that, and presumably the high bit is sometimes set
because the handle is generated with a PRNG.

Nevertheless, I see no point in using the wrong macro.  I'll plan on
committing/back-patching this shortly.

-- 
nathan



Commits

  1. Use correct DatumGet*() function in test_shm_mq_main().