From: Daniel Thompson <daniel.thompson@linaro.org>
To: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
Arnd Bergmann <arnd@arndb.de>,
Douglas Anderson <dianders@chromium.org>,
Jason Wessel <jason.wessel@windriver.com>,
kgdb-bugreport@lists.sourceforge.net
Cc: Daniel Thompson <daniel.thompson@linaro.org>,
linux-kernel@vger.kernel.org
Subject: [PATCH] kgdbts: Switch to do_sys_openat2() for breakpoint testing
Date: Thu, 25 Mar 2021 09:48:07 +0000 [thread overview]
Message-ID: <20210325094807.3546702-1-daniel.thompson@linaro.org> (raw)
Currently kgdbts can get stuck waiting for do_sys_open() to be called
in some of the current tests. This is because C compilers often
automatically inline this function, which is a very thin wrapper around
do_sys_openat2(), into some of its callers. gcc-10 does this on (at least)
both x86 and arm64.
We can fix the test suite by placing the breakpoints on do_sys_openat2()
instead since that isn't (currently) inlined. However do_sys_openat2() is
a static function so we cannot simply use an addressof. Since we are
testing debug machinery it is acceptable to use kallsyms to lookup a
suitable address because this is more or less what kdb does in the same
circumstances. Re-implement lookup_addr() to be based on kallsyms rather
than function pointers.
Signed-off-by: Daniel Thompson <daniel.thompson@linaro.org>
---
Notes:
So less than a week ago I said I had nothing pending for kgdbts.
That was entirely true when I said it... but then this came up.
drivers/misc/kgdbts.c | 48 ++++++++++++++++++++++++++-----------------
1 file changed, 29 insertions(+), 19 deletions(-)
diff --git a/drivers/misc/kgdbts.c b/drivers/misc/kgdbts.c
index 945701bce5536..2679374ca8361 100644
--- a/drivers/misc/kgdbts.c
+++ b/drivers/misc/kgdbts.c
@@ -92,6 +92,7 @@
#include <linux/kthread.h>
#include <linux/module.h>
#include <linux/sched/task.h>
+#include <linux/kallsyms.h>
#include <asm/sections.h>
@@ -200,21 +201,30 @@ static noinline void kgdbts_break_test(void)
v2printk("kgdbts: breakpoint complete\n");
}
-/* Lookup symbol info in the kernel */
+/*
+ * This is a cached wrapper for kallsyms_lookup_name().
+ *
+ * The cache is a big win for several tests. For example it more the doubles
+ * the cycles per second during the sys_open test. This is not theoretic,
+ * the performance improvement shows up at human scale, especially when
+ * testing using emulators.
+ *
+ * Obviously neither re-entrant nor thread-safe but that is OK since it
+ * can only be called from the debug trap (and therefore all other CPUs
+ * are halted).
+ */
static unsigned long lookup_addr(char *arg)
{
- unsigned long addr = 0;
-
- if (!strcmp(arg, "kgdbts_break_test"))
- addr = (unsigned long)kgdbts_break_test;
- else if (!strcmp(arg, "sys_open"))
- addr = (unsigned long)do_sys_open;
- else if (!strcmp(arg, "kernel_clone"))
- addr = (unsigned long)kernel_clone;
- else if (!strcmp(arg, "hw_break_val"))
- addr = (unsigned long)&hw_break_val;
- addr = (unsigned long) dereference_function_descriptor((void *)addr);
- return addr;
+ static char cached_arg[KSYM_NAME_LEN];
+ static unsigned long cached_addr;
+
+ if (strcmp(arg, cached_arg)) {
+ strscpy(cached_arg, arg, KSYM_NAME_LEN);
+ cached_addr = kallsyms_lookup_name(arg);
+ }
+
+ return (unsigned long)dereference_function_descriptor(
+ (void *)cached_addr);
}
static void break_helper(char *bp_type, char *arg, unsigned long vaddr)
@@ -310,7 +320,7 @@ static int check_and_rewind_pc(char *put_str, char *arg)
if (arch_needs_sstep_emulation && sstep_addr &&
ip + offset == sstep_addr &&
- ((!strcmp(arg, "sys_open") || !strcmp(arg, "kernel_clone")))) {
+ ((!strcmp(arg, "do_sys_openat2") || !strcmp(arg, "kernel_clone")))) {
/* This is special case for emulated single step */
v2printk("Emul: rewind hit single step bp\n");
restart_from_top_after_write = 1;
@@ -619,14 +629,14 @@ static struct test_struct do_kernel_clone_test[] = {
*/
static struct test_struct sys_open_test[] = {
{ "?", "S0*" }, /* Clear break points */
- { "sys_open", "OK", sw_break, }, /* set sw breakpoint */
+ { "do_sys_openat2", "OK", sw_break, }, /* set sw breakpoint */
{ "c", "T0*", NULL, get_thread_id_continue }, /* Continue */
- { "sys_open", "OK", sw_rem_break }, /*remove breakpoint */
- { "g", "sys_open", NULL, check_and_rewind_pc }, /* check location */
+ { "do_sys_openat2", "OK", sw_rem_break }, /*remove breakpoint */
+ { "g", "do_sys_openat2", NULL, check_and_rewind_pc }, /* check location */
{ "write", "OK", write_regs, emul_reset }, /* Write registers */
{ "s", "T0*", emul_sstep_get, emul_sstep_put }, /* Single step */
- { "g", "sys_open", NULL, check_single_step },
- { "sys_open", "OK", sw_break, }, /* set sw breakpoint */
+ { "g", "do_sys_openat2", NULL, check_single_step },
+ { "do_sys_openat2", "OK", sw_break, }, /* set sw breakpoint */
{ "7", "T0*", skip_back_repeat_test }, /* Loop based on repeat_test */
{ "D", "OK", NULL, final_ack_set }, /* detach and unregister I/O */
{ "", "", get_cont_catch, put_cont_catch },
base-commit: 1e28eed17697bcf343c6743f0028cc3b5dd88bf0
--
2.29.2
next reply other threads:[~2021-03-25 9:49 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-03-25 9:48 Daniel Thompson [this message]
2021-03-28 12:40 ` Greg Kroah-Hartman
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=20210325094807.3546702-1-daniel.thompson@linaro.org \
--to=daniel.thompson@linaro.org \
--cc=arnd@arndb.de \
--cc=dianders@chromium.org \
--cc=gregkh@linuxfoundation.org \
--cc=jason.wessel@windriver.com \
--cc=kgdb-bugreport@lists.sourceforge.net \
--cc=linux-kernel@vger.kernel.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®