[PATCH 0/2] dyndbg: fix incorrect mod_ct value in dynamic_debug_init()

[PATCH 0/2] dyndbg: fix incorrect mod_ct value in dynamic_debug_init()

[Yuntao Wang]

Fix and clean up the dynamic_debug_init() function.

Yuntao Wang (2):
  dyndbg: fix incorrect mod_ct value in dynamic_debug_init()
  dyndbg: clean up dynamic_debug_init() to improve readability

 lib/dynamic_debug.c | 15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)

--
2.55.0

[PATCH 1/2] dyndbg: fix incorrect mod_ct value in dynamic_debug_init()

[Yuntao Wang]

Suppose all `struct _ddebug` instances belong to the same module, mod_ct
should be 1, but it is currently 0.

Fix it.

Signed-off-by: Yuntao Wang <yuntao.wang@linux.dev>
---
 lib/dynamic_debug.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/lib/dynamic_debug.c b/lib/dynamic_debug.c
index 18a71a9108d3..16fad5454d6a 100644
--- a/lib/dynamic_debug.c
+++ b/lib/dynamic_debug.c
@@ -1456,6 +1456,8 @@ static int __init dynamic_debug_init(void)
 			iter_mod_start = iter;
 		}
 	}
+
+	mod_ct++;
 	di.num_descs = mod_sites;
 	di.descs = iter_mod_start;
 	ret = ddebug_add_module(&di, modname);
-- 
2.55.0

[PATCH 2/2] dyndbg: clean up dynamic_debug_init() to improve readability

[Yuntao Wang]

Keep variable assignments in the same order throughout the function to
make the code easier to follow.

No functional changes.

Signed-off-by: Yuntao Wang <yuntao.wang@linux.dev>
---
 lib/dynamic_debug.c | 15 ++++++++-------
 1 file changed, 8 insertions(+), 7 deletions(-)

diff --git a/lib/dynamic_debug.c b/lib/dynamic_debug.c
index 16fad5454d6a..49334d1aa4b3 100644
--- a/lib/dynamic_debug.c
+++ b/lib/dynamic_debug.c
@@ -1442,28 +1442,29 @@ static int __init dynamic_debug_init(void)
 	i = mod_sites = mod_ct = 0;
 
 	for (; iter < __stop___dyndbg; iter++, i++, mod_sites++) {
-
 		if (strcmp(modname, iter->modname)) {
-			mod_ct++;
-			di.num_descs = mod_sites;
 			di.descs = iter_mod_start;
+			di.num_descs = mod_sites;
 			ret = ddebug_add_module(&di, modname);
 			if (ret)
 				goto out_err;
 
-			mod_sites = 0;
-			modname = iter->modname;
+			mod_ct++;
+
 			iter_mod_start = iter;
+			modname = iter->modname;
+			mod_sites = 0;
 		}
 	}
 
-	mod_ct++;
-	di.num_descs = mod_sites;
 	di.descs = iter_mod_start;
+	di.num_descs = mod_sites;
 	ret = ddebug_add_module(&di, modname);
 	if (ret)
 		goto out_err;
 
+	mod_ct++;
+
 	ddebug_init_success = 1;
 	vpr_info("%d prdebugs in %d modules, %d KiB in ddebug tables, %d kiB in __dyndbg section\n",
 		 i, mod_ct, (int)((mod_ct * sizeof(struct ddebug_table)) >> 10),
-- 
2.55.0

Re: [PATCH 0/2] dyndbg: fix incorrect mod_ct value in dynamic_debug_init()

[Yuntao Wang]

Hi all,

Just a gentle ping on this patch series. Any chance someone could
take a look and review it?

I'd really appreciate it.

Thanks,
Yuntao

Re: [PATCH 1/2] dyndbg: fix incorrect mod_ct value in dynamic_debug_init()

[Andrew Morton]

On Tue, 11 Aug 2026 20:18:30 +0800 Yuntao Wang <yuntao.wang@linux.dev> wrote:

> Suppose all `struct _ddebug` instances belong to the same module, mod_ct
> should be 1, but it is currently 0.

Oh.  Why.  Does

		if (strcmp(modname, iter->modname)) {

not evaluate to true in this situation?

ie, more details please.

> Fix it.

What are the userspace-visible runtime effects of this bug?

Thanks.

> --- a/lib/dynamic_debug.c
> +++ b/lib/dynamic_debug.c
> @@ -1456,6 +1456,8 @@ static int __init dynamic_debug_init(void)
>  			iter_mod_start = iter;
>  		}
>  	}
> +
> +	mod_ct++;
>  	di.num_descs = mod_sites;
>  	di.descs = iter_mod_start;
>  	ret = ddebug_add_module(&di, modname);
> -- 
> 2.55.0
> 

Re: [PATCH 1/2] dyndbg: fix incorrect mod_ct value in dynamic_debug_init()

[Yuntao Wang]

On Sat, 29 Aug 2026 20:23:06 -0700, Andrew Morton <akpm@linux-foundation.org> wrote:

> On Tue, 11 Aug 2026 20:18:30 +0800 Yuntao Wang <yuntao.wang@linux.dev> wrote:
>
> > Suppose all `struct _ddebug` instances belong to the same module, mod_ct
> > should be 1, but it is currently 0.
>
> Oh.  Why.  Does
>
> 		if (strcmp(modname, iter->modname)) {
>
> not evaluate to true in this situation?
>
> ie, more details please.
>
> > Fix it.
>
> What are the userspace-visible runtime effects of this bug?
>
> Thanks.

Hi Andrew,

mod_ct is incremented only when iter->modname changes, i.e. when the loop
encounters the first _ddebug entry of a new module:

if (strcmp(modname, iter->modname)) {
	mod_ct++;
	...
}

If all _ddebug entries belong to the same module, strcmp() never returns
nonzero, so mod_ct remains 0.

However, the last (and in this case only) module is added after the loop:

di.num_descs = mod_sites;
di.descs = iter_mod_start;
ret = ddebug_add_module(&di, modname);

Thus, mod_ct should be incremented before adding this final module.

The bug only affects the diagnostic message printed by vpr_info():

"%d prdebugs in %d modules, ..."

It reports one fewer module than the actual number of modules. There is no
userspace-visible runtime effect; the dynamic debug tables themselves are
initialized correctly.

Thanks,
Yuntao

> > --- a/lib/dynamic_debug.c
> > +++ b/lib/dynamic_debug.c
> > @@ -1456,6 +1456,8 @@ static int __init dynamic_debug_init(void)
> >  			iter_mod_start = iter;
> >  		}
> >  	}
> > +
> > +	mod_ct++;
> >  	di.num_descs = mod_sites;
> >  	di.descs = iter_mod_start;
> >  	ret = ddebug_add_module(&di, modname);
> > --
> > 2.55.0
> >