[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAGudoHETsUtwR=KJpMCEfbZazxv-FbYgMSe_Lk4MV0CVg0h9ow@mail.gmail.com>
Date: Thu, 15 Aug 2024 21:07:26 +0200
From: Mateusz Guzik <mjguzik@...il.com>
To: Jann Horn <jannh@...gle.com>
Cc: Suren Baghdasaryan <surenb@...gle.com>, Christian Brauner <brauner@...nel.org>,
Andrii Nakryiko <andrii.nakryiko@...il.com>, Andrii Nakryiko <andrii@...nel.org>,
linux-trace-kernel@...r.kernel.org, peterz@...radead.org, oleg@...hat.com,
rostedt@...dmis.org, mhiramat@...nel.org, bpf@...r.kernel.org,
linux-kernel@...r.kernel.org, jolsa@...nel.org, paulmck@...nel.org,
willy@...radead.org, akpm@...ux-foundation.org, linux-mm@...ck.org
Subject: Re: [PATCH RFC v3 13/13] uprobes: add speculative lockless VMA to
inode resolution
On Thu, Aug 15, 2024 at 8:58 PM Jann Horn <jannh@...gle.com> wrote:
> Stupid question: Is this uprobe stuff actually such a hot codepath
> that it makes sense to optimize it to be faster than the page fault
> path?
>
That's what I implicitly asked, hoping a down_read on vma would do it,
but Andrii claims multiple parallel lookups on the same vma are a
problem.
Even so, I suspect something *simple* is doable here which avoids any
writes to vmas and does not need the mm-wide sequence counter. It may
be requirements are lax enough that merely observing some state is the
same before and after uprobe lookup will be sufficient, or maybe some
other hackery is viable without messing with fences in
vma_start_write.
--
Mateusz Guzik <mjguzik gmail.com>
Powered by blists - more mailing lists