Re: free and errno, was Re: [PATCH] apply: replace mksnpath() with a mkpathdup() call

From: Jeff King <peff@peff.net>
Date: Apr 5, 2024, 17:41 UTC
Message-ID: <20240405174136.GC2529133@coredump.intra.peff.net>
In-Reply-To: <20240405173517.GA2529133@coredump.intra.peff.net>

On Fri, Apr 05, 2024 at 01:35:17PM -0400, Jeff King wrote:

Show 6 quoted lines

> With the patch below you can see both in action:
> 
>   - hyperfine seems to show the git_free() version as ~1% slower to run
>     "git log" (which I picked as something that probably does a
>     reasonable number of mallocs). Frankly, I'm still suspicious that it
>     could have such an impact. Maybe inlining git_free() would help?

This may have been noise. Weirdly it seemed consistent from run to run using hyperfine. But switching my CPU governor from "powersave" to "performance", I got a run where the new code is fast.

-Peff