[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20210218193508.mw6ugmh6bjlplbwe@treble>
Date: Thu, 18 Feb 2021 13:35:08 -0600
From: Josh Poimboeuf <jpoimboe@...hat.com>
To: Evgenii Shatokhin <eshatokhin@...tuozzo.com>
Cc: Masami Hiramatsu <mhiramat@...nel.org>,
Arnaldo Carvalho de Melo <acme@...nel.org>,
Kristen Carlson Accardi <kristen@...ux.intel.com>,
live-patching@...r.kernel.org,
Peter Zijlstra <peterz@...radead.org>,
Ingo Molnar <mingo@...hat.com>, linux-kernel@...r.kernel.org,
Konstantin Khorenko <khorenko@...tuozzo.com>
Subject: Re: 'perf probe' and symbols from .text.<something>
On Thu, Feb 18, 2021 at 08:09:17PM +0300, Evgenii Shatokhin wrote:
> Is there a way to allow probing of functions in .text.<something> ?
>
> Of course, one could place probes using absolute addresses of the functions
> but that would be less convenient.
>
> This also affects many livepatch modules where the kernel code can be
> compiled with -ffunction-sections and each function may end up in a separate
> section .text.<function_name>. 'perf probe' cannot be used there, except
> with the absolute addresses.
>
> Moreover, if FGKASLR patches are merged (https://lwn.net/Articles/832434/)
> and the kernel is built with FGKASLR enabled, -ffunction-sections will be
> used too. 'perf probe' will be unable to see the kernel functions then.
A hack fix like the below would probably work, but as you pointed out,
FGKASLR is going to be a problem. I suspect the proper fix is for perf
to learn how to deal with multiple executable ELF sections.
diff --git a/scripts/module.lds.S b/scripts/module.lds.S
index 69b9b71a6a47..0c522a87f6ce 100644
--- a/scripts/module.lds.S
+++ b/scripts/module.lds.S
@@ -4,6 +4,10 @@
* combine them automatically.
*/
SECTIONS {
+ .text : {
+ *(.text)
+ *(.text.*)
+ }
/DISCARD/ : {
*(.discard)
*(.discard.*)
Powered by blists - more mailing lists