From: Miroslav Benes <mbenes@suse.cz>
Date: Thu, 9 Sep 2021 13:53:35 +0200 (CEST)

Hi,

Hi!

On Tue, 31 Aug 2021, Alexander Lobakin wrote:

quoted

From: Kristen Carlson Accardi <redacted>

If any type of function granular randomization is enabled, the sympos
algorithm will fail, as it will be impossible to resolve symbols when
there are duplicates using the previous symbol position.

Override the value of sympos to always be zero if fgkaslr is enabled for
either the core kernel or modules, forcing the algorithm
to require that only unique symbols are allowed to be patched.

Signed-off-by: Kristen Carlson Accardi <redacted>
Signed-off-by: Alexander Lobakin <redacted>
---
 kernel/livepatch/core.c | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/kernel/livepatch/core.c b/kernel/livepatch/core.c
index 335d988bd811..852bbfa9da7b 100644
--- a/kernel/livepatch/core.c
+++ b/kernel/livepatch/core.c

@@ -169,6 +169,17 @@ static int klp_find_object_symbol(const char *objname, const char *name,
 	else
 		kallsyms_on_each_symbol(klp_find_callback, &args);
 
+	/*
+	 * If any type of function granular randomization is enabled, it
+	 * will be impossible to resolve symbols when there are duplicates
+	 * using the previous symbol position (i.e. sympos != 0). Override
+	 * the value of sympos to always be zero in this case. This will
+	 * force the algorithm to require that only unique symbols are
+	 * allowed to be patched.
+	 */
+	if (IS_ENABLED(CONFIG_FG_KASLR))
+		sympos = 0;
+

I ran the live patching tests and no problem occurred, which is great. We 
do not have a test for old_sympos, which makes the testing less telling, 
but at least nothing blows up with the section randomization itself.

Great, thanks!

However, I want to reiterate what I wrote for the same patch in v5 
series.

The above hunk should work, but I wonder if we should make it more 
explicit. With the change the user will get the error with "unresolvable 
ambiguity for symbol..." if they specify sympos and the symbol is not 
unique. It could confuse them.

So, how about it making it something like

if (IS_ENABLED(CONFIG_FG_KASLR) || IS_ENABLED(CONFIG_MODULE_FG_KASLR))
        if (sympos) {
                pr_err("fgkaslr is enabled, specifying sympos for symbol '%s' in object '%s' does not work.\n",
                        name, objname);
                *addr = 0;
                return -EINVAL;
        }

? (there could be goto to the error out at the end of the function to 
save copy-pasting).

In that case, if sympos is not specified, the user will get the message 
which matches the reality. If the user specifies it, they will get the 
error in case of fgkaslr (no matter if the symbol is found or not).

Not familiar with livepatching unfortunately, hope Kristen and/or
Kees will comment on this. Looks fine for me anyways.

What do you think?

Miroslav

Thanks,
Al