GH-145278: also filter mmap2 in strace_helper.filter_memory

[FFY00]

• Issue: Freeze modules used during startup #145278

[bedevere-bot]

🤖 New build scheduled with the buildbot fleet by @FFY00 for commit 1880772 🤖

Results will be shown at:

https://buildbot.python.org/all/#/grid?branch=refs%2Fpull%2F148648%2Fmerge

If you want to schedule another build, you need to add the 🔨 test-with-buildbots label again.

[vstinner]

LGTM

[vstinner]

We should wait for "ARM Raspbian 3.x" buildbot worker to report if tests pass or not. Right now, there are two queued job on "ARM Raspbian PR": https://buildbot.python.org/#/builders/1062.

[FFY00]

Yep, I am monitoring it to see if this change actually fixes the buildbot failure 👍

[FFY00]

Raspbian seems to be passing now.

https://buildbot.python.org/all/#/builders/1062/builds/1357

I'll wait for the rest of the buildbots to finish, and then I'll merge.

[cmaloney]

Also would be happy with call.syscall.startswith('mmap') here as a guard against mmap3 / future extensions as well.

[encukou]

PR Buildbots are green; I'll merge.

startswith can go in the next PR -- third time's the charm, right?

[cmaloney]

This is more like 5th ><. There were a couple reverts getting bots clean to land the strace testing infrastructure...

If there is a more reliable way to check for accidental additions of stat or read system calls when reading a file am all ears. At least 3 times before the strace test people optimized and then more calls were accidentally added...

[encukou]

Well, the best way to classify accidental vs. intended additions is, still, human review. Can we make that easier? Perhaps change an error message to point to the place to fix?