Message-ID: <00e2f579-99d8-4b80-94c0-745612ac2ff5@gmail.com>
Date: Mon, 23 Feb 2026 14:31:34 -0600
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: [PATCH] Add Windows support for backtrace_functions (MSVC only)
To: =?UTF-8?Q?=C3=81lvaro_Herrera?= <alvherre@kurilemu.de>
Cc: Euler Taveira <euler@eulerto.com>,
 Jakub Wartak <jakub.wartak@enterprisedb.com>,
 Michael Paquier <michael@paquier.xyz>,
 pgsql-hackers <pgsql-hackers@postgresql.org>
References: <202602232005.2cmbj5ougsfv@alvherre.pgsql>
Content-Language: en-US
From: Bryan Green <dbryan.green@gmail.com>
In-Reply-To: <202602232005.2cmbj5ougsfv@alvherre.pgsql>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Archived-At: 
 <https://www.postgresql.org/message-id/00e2f579-99d8-4b80-94c0-745612ac2ff5%40gmail.com>
Precedence: bulk

On 2/23/2026 2:16 PM, Álvaro Herrera wrote:
> Hello, thanks for following up.
> 
> However, I was going over the commit message one final time after your
> reply, and I noticed that SymInitialize()'s documentation says quite
> explicitly:
> 
> "Do not use the handle returned by GetCurrentProcess"
> https://learn.microsoft.com/en-us/windows/win32/api/dbghelp/nf-dbghelp-syminitialize
> 
> which is exactly what we do with this patch.
> 
> I think we should DuplicateHandle() on the GetCurrentProcess() handle
> (which then obviously needs CloseHandle afterwards).  Would you agree?
> 
> Here's the whole thing again with edited commit message and some trivial
> stylistic changes.
> 
> 
> I was amused to discover that Solaris, macOS and FreeBSD all copied the
> backtrace() interface from glibc, and the only platform in the buildfarm
> other than Windows that seems to lack support for backtraces is Alpine
> Linux.
> 
> Regards
> 
Ouch.  Can't believe I missed that.  Yeah, that is a no go.  We could
use openprocess with the getcurrentprocessid.  Let me look this over.
And yes, regardless of which way we do it we will need a CloseHandle.

-- 
Bryan Green
EDB: https://www.enterprisedb.com