[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <1651166416.afl3n7tts7.naveen@linux.ibm.com>
Date: Thu, 28 Apr 2022 22:54:38 +0530
From: "Naveen N. Rao" <naveen.n.rao@...ux.vnet.ibm.com>
To: Steven Rostedt <rostedt@...dmis.org>
Cc: linux-kernel@...r.kernel.org, linuxppc-dev@...ts.ozlabs.org,
llvm@...ts.linux.dev, Michael Ellerman <mpe@...erman.id.au>,
Nathan Chancellor <nathan@...nel.org>,
Nick Desaulniers <ndesaulniers@...gle.com>
Subject: Re: [PATCH 2/2] recordmcount: Handle sections with no non-weak
symbols
Steven Rostedt wrote:
> On Thu, 28 Apr 2022 13:15:22 +0530
> "Naveen N. Rao" <naveen.n.rao@...ux.vnet.ibm.com> wrote:
>
>> Indeed, plain old -pg will be a problem. I'm not sure there is a generic
>> way to address this. I suppose architectures will have to validate the
>> mcount locations, something like this?
>
> Perhaps another solution is to make the mcount locations after the linking
> is done. The main downside to that is that it takes time to go over the
> entire vmlinux, and will slow down a compile that only modified a couple of
> files.
Yes, and I think that is also very useful with LTO. So, that would be a
good one to consider in the longer term.
For now, I have posted a v2 of this series with your comments addressed.
It is working well in my tests on powerpc in the different
configurations, including the older elf abi v1 that uses -pg. If it
looks ok to you, we can use this approach for now.
Thanks,
Naveen
Powered by blists - more mailing lists