- This series fixes -Wformat-truncation warnings in staging drivers found with W=1 builds: - greybus: prefix device id to widget/control names safely - vc04_services/bcm2835-camera: compose V4L2 bus_info safely Both changes use scnprintf() for the prefix and strscpy() for the remainder. No functional change intended.
Masaharu Noguchi (2): staging: greybus: audio_topology: avoid -Wformat-truncation staging: vc04_services: bcm2835-camera: avoid -Wformat-truncation
drivers/staging/greybus/audio_topology.c | 11 ++++++----- .../vc04_services/bcm2835-camera/bcm2835-camera.c | 5 +++-- 2 files changed, 9 insertions(+), 7 deletions(-)
- Fix -Wformat-truncation when prefixing device id to widget/control names. - Write the prefix with scnprintf() and copy the remainder with strscpy(). - This avoids potential truncation and satisfies W=1 builds in drivers/staging/greybus.
Signed-off-by: Masaharu Noguchi nogunix@gmail.com --- drivers/staging/greybus/audio_topology.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-)
diff --git a/drivers/staging/greybus/audio_topology.c b/drivers/staging/greybus/audio_topology.c index 6ca938dca4fd..5bf8b5e29dd8 100644 --- a/drivers/staging/greybus/audio_topology.c +++ b/drivers/staging/greybus/audio_topology.c @@ -1013,7 +1013,7 @@ static int gbaudio_tplg_create_widget(struct gbaudio_module_info *module, struct snd_soc_dapm_widget *dw, struct gb_audio_widget *w, int *w_size) { - int i, ret, csize; + int i, ret, csize, n; struct snd_kcontrol_new *widget_kctls; struct gb_audio_control *curr; struct gbaudio_control *control, *_control; @@ -1087,7 +1087,8 @@ static int gbaudio_tplg_create_widget(struct gbaudio_module_info *module,
/* Prefix dev_id to widget control_name */ strscpy(temp_name, w->name, sizeof(temp_name)); - snprintf(w->name, sizeof(w->name), "GB %d %s", module->dev_id, temp_name); + n = scnprintf(w->name, sizeof(w->name), "GB %d ", module->dev_id); + strscpy(w->name + n, temp_name, sizeof(w->name) - n);
switch (w->type) { case snd_soc_dapm_spk: @@ -1138,7 +1139,7 @@ static int gbaudio_tplg_create_widget(struct gbaudio_module_info *module, static int gbaudio_tplg_process_kcontrols(struct gbaudio_module_info *module, struct gb_audio_control *controls) { - int i, csize, ret; + int i, csize, ret, n; struct snd_kcontrol_new *dapm_kctls; struct gb_audio_control *curr; struct gbaudio_control *control, *_control; @@ -1169,8 +1170,8 @@ static int gbaudio_tplg_process_kcontrols(struct gbaudio_module_info *module, control->id = curr->id; /* Prefix dev_id to widget_name */ strscpy(temp_name, curr->name, sizeof(temp_name)); - snprintf(curr->name, sizeof(curr->name), "GB %d %s", module->dev_id, - temp_name); + n = scnprintf(curr->name, sizeof(curr->name), "GB %d ", module->dev_id); + strscpy(curr->name + n, temp_name, sizeof(curr->name) - n); control->name = curr->name; if (curr->info.type == GB_AUDIO_CTL_ELEM_TYPE_ENUMERATED) { struct gb_audio_enumerated *gbenum =
On Sun, Aug 31, 2025 at 02:38:49AM +0900, Masaharu Noguchi wrote:
- Fix -Wformat-truncation when prefixing device id to widget/control names. - Write the prefix with scnprintf() and copy the remainder with strscpy(). - This avoids potential truncation and satisfies W=1 builds in drivers/staging/greybus.Signed-off-by: Masaharu Noguchi nogunix@gmail.com
drivers/staging/greybus/audio_topology.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-)
diff --git a/drivers/staging/greybus/audio_topology.c b/drivers/staging/greybus/audio_topology.c index 6ca938dca4fd..5bf8b5e29dd8 100644 --- a/drivers/staging/greybus/audio_topology.c +++ b/drivers/staging/greybus/audio_topology.c @@ -1013,7 +1013,7 @@ static int gbaudio_tplg_create_widget(struct gbaudio_module_info *module, struct snd_soc_dapm_widget *dw, struct gb_audio_widget *w, int *w_size) {
- int i, ret, csize;
- int i, ret, csize, n; struct snd_kcontrol_new *widget_kctls; struct gb_audio_control *curr; struct gbaudio_control *control, *_control;
@@ -1087,7 +1087,8 @@ static int gbaudio_tplg_create_widget(struct gbaudio_module_info *module, /* Prefix dev_id to widget control_name */ strscpy(temp_name, w->name, sizeof(temp_name));
- snprintf(w->name, sizeof(w->name), "GB %d %s", module->dev_id, temp_name);
- n = scnprintf(w->name, sizeof(w->name), "GB %d ", module->dev_id);
- strscpy(w->name + n, temp_name, sizeof(w->name) - n);
The strscpy() doesn't make sense. If the string doesn't fit then you can't *force* it to fit with strscpy(). :P
Here we're taking a buffer w->name and adding a prefix at the beginning of the exact same buffer. Obviously the result is going to be larger. The W=1 warning doesn't add any new information...
I feel like these W=1 warnings about string truncation are rarely useful. These limits are not normally thought out that deeply. Poeople just say, "This is probably something like a company name. Let's say that a company name is probably 48 characters long." But really very few company names are that long. It's just a rough estimate.
And really it's not the worst thing if these strings are truncated. Kernel messages are mostly error messages. We'll still be able to debug the crash even if the last couple characters in a really long name are chopped off.
regards, dan carpenter
Thanks for the detailed explanation. I understand your point regarding W=1 warnings and string truncation. Based on this, I will drop the previous patch and leave the code as is.
Best regards, Masaharu Noguchi
- Fix -Wformat-truncation in vidioc_querycap() when composing bus_info. - Use scnprintf() for the prefix and strscpy() for the remainder.
Signed-off-by: Masaharu Noguchi nogunix@gmail.com --- .../staging/vc04_services/bcm2835-camera/bcm2835-camera.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/drivers/staging/vc04_services/bcm2835-camera/bcm2835-camera.c b/drivers/staging/vc04_services/bcm2835-camera/bcm2835-camera.c index fa7ea4ca4c36..c2788659af12 100644 --- a/drivers/staging/vc04_services/bcm2835-camera/bcm2835-camera.c +++ b/drivers/staging/vc04_services/bcm2835-camera/bcm2835-camera.c @@ -874,13 +874,14 @@ static int vidioc_querycap(struct file *file, void *priv, struct bcm2835_mmal_dev *dev = video_drvdata(file); u32 major; u32 minor; + int n;
vchiq_mmal_version(dev->instance, &major, &minor);
strscpy(cap->driver, "bcm2835 mmal", sizeof(cap->driver)); snprintf((char *)cap->card, sizeof(cap->card), "mmal service %d.%d", major, minor); - - snprintf((char *)cap->bus_info, sizeof(cap->bus_info), "platform:%s", dev->v4l2_dev.name); + n = scnprintf((char *)cap->bus_info, sizeof(cap->bus_info), "platform:"); + strscpy((char *)cap->bus_info + n, dev->v4l2_dev.name, sizeof(cap->bus_info) - n); return 0; }
On Sun, Aug 31, 2025 at 02:38:50AM +0900, Masaharu Noguchi wrote:
- Fix -Wformat-truncation in vidioc_querycap() when composing bus_info. - Use scnprintf() for the prefix and strscpy() for the remainder.Signed-off-by: Masaharu Noguchi nogunix@gmail.com
.../staging/vc04_services/bcm2835-camera/bcm2835-camera.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/drivers/staging/vc04_services/bcm2835-camera/bcm2835-camera.c b/drivers/staging/vc04_services/bcm2835-camera/bcm2835-camera.c index fa7ea4ca4c36..c2788659af12 100644 --- a/drivers/staging/vc04_services/bcm2835-camera/bcm2835-camera.c +++ b/drivers/staging/vc04_services/bcm2835-camera/bcm2835-camera.c @@ -874,13 +874,14 @@ static int vidioc_querycap(struct file *file, void *priv, struct bcm2835_mmal_dev *dev = video_drvdata(file); u32 major; u32 minor;
- int n;
vchiq_mmal_version(dev->instance, &major, &minor); strscpy(cap->driver, "bcm2835 mmal", sizeof(cap->driver)); snprintf((char *)cap->card, sizeof(cap->card), "mmal service %d.%d", major, minor);
- snprintf((char *)cap->bus_info, sizeof(cap->bus_info), "platform:%s", dev->v4l2_dev.name);
- n = scnprintf((char *)cap->bus_info, sizeof(cap->bus_info), "platform:");
- strscpy((char *)cap->bus_info + n, dev->v4l2_dev.name, sizeof(cap->bus_info) - n);
The fact that we need such a complicated construct is a sign that there's something wrong in our APIs. The above code is too complicated for what it does, making it less readable, more difficult to maintain, and more bug-prone. I don't know if we need yet another sprintf variant in the kernel, or something else, but this doens't seem to be what we need.
return 0; }
Laurent Pinchart,
Thanks for your feedback.
I see your point regarding the complexity of the current construct. I’ll take some time to consider the best approach and follow up once I have a clearer picture.
Masaharu Noguchi
On Sat, Aug 30, 2025 at 11:29:23PM +0200, Laurent Pinchart wrote:
The fact that we need such a complicated construct is a sign that there's something wrong in our APIs. The above code is too complicated for what it does, making it less readable, more difficult to maintain, and more bug-prone. I don't know if we need yet another sprintf variant in the kernel, or something else, but this doens't seem to be what we need.
}
-- Regards,
Laurent Pinchart
Thanks for the explanation. I understand now that this patch is not needed. Please drop it.
Best regards, Masaharu Noguchi
-
Dan Carpenter -
Laurent Pinchart -
Masaharu Noguchi