From: Josh Poimboeuf <jpoimboe@redhat.com>
To: Allan Xavier <allan.x.xavier@oracle.com>
Cc: Peter Zijlstra <peterz@infradead.org>, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] objtool: Fix GCC 8 cold function processing without -freorder-functions
Date: Tue, 26 Jun 2018 13:44:05 -0500 [thread overview]
Message-ID: <20180626184405.hsbuuq5eo7qzicoi@treble> (raw)
In-Reply-To: <3a9a1cfa-4357-264e-3333-220a7f83a51d@oracle.com>
On Tue, Jun 26, 2018 at 07:31:18PM +0100, Allan Xavier wrote:
> Hi Peter,
>
> On 26/06/18 17:43, Peter Zijlstra wrote:
> > On Tue, Jun 26, 2018 at 05:20:45PM +0100, Allan Xavier wrote:
> >> 0000000000000500 g F .text 0000000000000034 nmi_panic
> >> 0000000000000528 l F .text 000000000000000c nmi_panic.cold.7
> >>
> >> This doesn't happen with -freorder-functions in the first example as the
> >> symbols don't overlap.
> >
> > Urgh and I don't suppose we can 'fix' the overlap in read_symbols() ?
> >
>
> It should be fixable in read_symbols too if you don't mind sym->len not being
> the same as sym->st_size in the ELF.
>
> Is there a particular concern you're trying to address by having the logic there
> instead?
>
> Will have a look into reworking the patch in any case.
Thanks for the patch and the excellent problem description. I think the
concern is that the overlap might cause other unforeseen issues (now or
in the future). How about something like this? (The diff also includes
a cleanup to reduce the indent level.)
diff --git a/tools/objtool/elf.c b/tools/objtool/elf.c
index 4e60e105583e..36518393a960 100644
--- a/tools/objtool/elf.c
+++ b/tools/objtool/elf.c
@@ -302,19 +302,32 @@ static int read_symbols(struct elf *elf)
continue;
sym->pfunc = sym->cfunc = sym;
coldstr = strstr(sym->name, ".cold.");
- if (coldstr) {
- coldstr[0] = '\0';
- pfunc = find_symbol_by_name(elf, sym->name);
- coldstr[0] = '.';
-
- if (!pfunc) {
- WARN("%s(): can't find parent function",
- sym->name);
- goto err;
- }
-
- sym->pfunc = pfunc;
- pfunc->cfunc = sym;
+ if (!coldstr)
+ continue;
+
+ coldstr[0] = '\0';
+ pfunc = find_symbol_by_name(elf, sym->name);
+ coldstr[0] = '.';
+
+ if (!pfunc) {
+ WARN("%s(): can't find parent function",
+ sym->name);
+ goto err;
+ }
+
+ sym->pfunc = pfunc;
+ pfunc->cfunc = sym;
+
+ /*
+ * Unfortunately, -fnoreorder-functions puts the child
+ * inside the parent. Remove the overlap so we can
+ * have sane assumptions.
+ */
+ if (sym->sec == pfunc->sec &&
+ sym->offset >= pfunc->offset &&
+ sym->offset < pfunc->offset + pfunc->len &&
+ sym->offset + sym->len == pfunc->offset + pfunc->len) {
+ pfunc->len -= sym->len;
}
}
}
next prev parent reply other threads:[~2018-06-26 18:44 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-06-26 16:20 Allan Xavier
2018-06-26 16:43 ` Peter Zijlstra
2018-06-26 18:31 ` Allan Xavier
2018-06-26 18:44 ` Josh Poimboeuf [this message]
2018-06-26 19:43 ` Peter Zijlstra
2018-06-26 21:07 ` Josh Poimboeuf
2018-06-27 9:35 ` Allan Xavier
2018-06-27 15:15 ` Josh Poimboeuf
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20180626184405.hsbuuq5eo7qzicoi@treble \
--to=jpoimboe@redhat.com \
--cc=allan.x.xavier@oracle.com \
--cc=linux-kernel@vger.kernel.org \
--cc=peterz@infradead.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®