drivers/soundwire/debugfs.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
In read_buffer_show(), allocate sufficient memory to pass on to
sysfs_emit(), which expects a buffer of size PAGE_SIZE.
Link: https://scan7.scan.coverity.com/#/project-view/53936/11354?selectedIssue=1648019
Fixes: 35323d8ab811 ("soundwire: replace scnprintf() with sysfs_emit() for sysfs consistency")
Signed-off-by: Sergio Perez Gonzalez <sperezglz@gmail.com>
---
drivers/soundwire/debugfs.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/drivers/soundwire/debugfs.c b/drivers/soundwire/debugfs.c
index 825f7abbad32..08fb1e29026e 100644
--- a/drivers/soundwire/debugfs.c
+++ b/drivers/soundwire/debugfs.c
@@ -306,12 +306,12 @@ static int cmd_go(void *data, u64 value)
DEFINE_DEBUGFS_ATTRIBUTE(cmd_go_fops, NULL,
cmd_go, "%llu\n");
-#define MAX_LINE_LEN 128
-
static int read_buffer_show(struct seq_file *s_file, void *data)
{
- char buf[MAX_LINE_LEN];
int i;
+ char *buf __free(kfree) = kzalloc(PAGE_SIZE, GFP_KERNEL);
+ if (!buf)
+ return -ENOMEM;
if (num_bytes == 0 || num_bytes > MAX_CMD_BYTES)
return -EINVAL;
--
2.43.0
On Thu, 24 Jul 2025 01:09:25 +0200, Sergio Perez Gonzalez wrote: > > In read_buffer_show(), allocate sufficient memory to pass on to > sysfs_emit(), which expects a buffer of size PAGE_SIZE. > > Link: https://scan7.scan.coverity.com/#/project-view/53936/11354?selectedIssue=1648019 > Fixes: 35323d8ab811 ("soundwire: replace scnprintf() with sysfs_emit() for sysfs consistency") > Signed-off-by: Sergio Perez Gonzalez <sperezglz@gmail.com> > --- > drivers/soundwire/debugfs.c | 6 +++--- > 1 file changed, 3 insertions(+), 3 deletions(-) > > diff --git a/drivers/soundwire/debugfs.c b/drivers/soundwire/debugfs.c > index 825f7abbad32..08fb1e29026e 100644 > --- a/drivers/soundwire/debugfs.c > +++ b/drivers/soundwire/debugfs.c > @@ -306,12 +306,12 @@ static int cmd_go(void *data, u64 value) > DEFINE_DEBUGFS_ATTRIBUTE(cmd_go_fops, NULL, > cmd_go, "%llu\n"); > > -#define MAX_LINE_LEN 128 > - > static int read_buffer_show(struct seq_file *s_file, void *data) > { > - char buf[MAX_LINE_LEN]; > int i; > + char *buf __free(kfree) = kzalloc(PAGE_SIZE, GFP_KERNEL); > + if (!buf) > + return -ENOMEM; IMO, it's better to scratch the whole previous change. It makes little sense to allocate a large buffer here just for the temporary formatting. Moreover, *_show() functions there take seq_file pointer, and you can just use seq_printf() directly. The sysfs/kobject show callback is with the a fixed PAGE_SIZE buffer, hence sysfs_emit() is useful, but in this case, it's a completely different story. thanks, Takashi
On 24-07-25, 09:19, Takashi Iwai wrote: > On Thu, 24 Jul 2025 01:09:25 +0200, > Sergio Perez Gonzalez wrote: > > > > In read_buffer_show(), allocate sufficient memory to pass on to > > sysfs_emit(), which expects a buffer of size PAGE_SIZE. > > > > Link: https://scan7.scan.coverity.com/#/project-view/53936/11354?selectedIssue=1648019 > > Fixes: 35323d8ab811 ("soundwire: replace scnprintf() with sysfs_emit() for sysfs consistency") > > Signed-off-by: Sergio Perez Gonzalez <sperezglz@gmail.com> > > --- > > drivers/soundwire/debugfs.c | 6 +++--- > > 1 file changed, 3 insertions(+), 3 deletions(-) > > > > diff --git a/drivers/soundwire/debugfs.c b/drivers/soundwire/debugfs.c > > index 825f7abbad32..08fb1e29026e 100644 > > --- a/drivers/soundwire/debugfs.c > > +++ b/drivers/soundwire/debugfs.c > > @@ -306,12 +306,12 @@ static int cmd_go(void *data, u64 value) > > DEFINE_DEBUGFS_ATTRIBUTE(cmd_go_fops, NULL, > > cmd_go, "%llu\n"); > > > > -#define MAX_LINE_LEN 128 > > - > > static int read_buffer_show(struct seq_file *s_file, void *data) > > { > > - char buf[MAX_LINE_LEN]; > > int i; > > + char *buf __free(kfree) = kzalloc(PAGE_SIZE, GFP_KERNEL); > > + if (!buf) > > + return -ENOMEM; > > IMO, it's better to scratch the whole previous change. > It makes little sense to allocate a large buffer here just for the > temporary formatting. > > Moreover, *_show() functions there take seq_file pointer, and you can > just use seq_printf() directly. The sysfs/kobject show callback is > with the a fixed PAGE_SIZE buffer, hence sysfs_emit() is useful, but > in this case, it's a completely different story. Agree, dropped now! -- ~Vinod
All right, then, this will be handled with a different approach. Thanks, Sergio El jue, 24 jul 2025 a la(s) 3:52 a.m., Vinod Koul (vkoul@kernel.org) escribió: > > On 24-07-25, 09:19, Takashi Iwai wrote: > > On Thu, 24 Jul 2025 01:09:25 +0200, > > Sergio Perez Gonzalez wrote: > > > > > > In read_buffer_show(), allocate sufficient memory to pass on to > > > sysfs_emit(), which expects a buffer of size PAGE_SIZE. > > > > > > Link: https://scan7.scan.coverity.com/#/project-view/53936/11354?selectedIssue=1648019 > > > Fixes: 35323d8ab811 ("soundwire: replace scnprintf() with sysfs_emit() for sysfs consistency") > > > Signed-off-by: Sergio Perez Gonzalez <sperezglz@gmail.com> > > > --- > > > drivers/soundwire/debugfs.c | 6 +++--- > > > 1 file changed, 3 insertions(+), 3 deletions(-) > > > > > > diff --git a/drivers/soundwire/debugfs.c b/drivers/soundwire/debugfs.c > > > index 825f7abbad32..08fb1e29026e 100644 > > > --- a/drivers/soundwire/debugfs.c > > > +++ b/drivers/soundwire/debugfs.c > > > @@ -306,12 +306,12 @@ static int cmd_go(void *data, u64 value) > > > DEFINE_DEBUGFS_ATTRIBUTE(cmd_go_fops, NULL, > > > cmd_go, "%llu\n"); > > > > > > -#define MAX_LINE_LEN 128 > > > - > > > static int read_buffer_show(struct seq_file *s_file, void *data) > > > { > > > - char buf[MAX_LINE_LEN]; > > > int i; > > > + char *buf __free(kfree) = kzalloc(PAGE_SIZE, GFP_KERNEL); > > > + if (!buf) > > > + return -ENOMEM; > > > > IMO, it's better to scratch the whole previous change. > > It makes little sense to allocate a large buffer here just for the > > temporary formatting. > > > > Moreover, *_show() functions there take seq_file pointer, and you can > > just use seq_printf() directly. The sysfs/kobject show callback is > > with the a fixed PAGE_SIZE buffer, hence sysfs_emit() is useful, but > > in this case, it's a completely different story. > > Agree, dropped now! > > -- > ~Vinod
© 2016 - 2025 Red Hat, Inc.