[PATCH v1 1/3] auxdisplay: lcd2s: Fix lcd2s_redefine_char() feature

[PATCH v1 1/3] auxdisplay: lcd2s: Fix lcd2s_redefine_char() feature

[Andy Shevchenko]

It seems that the lcd2s_redefine_char() has never been properly
tested. The buffer is filled by DEF_CUSTOM_CHAR command followed
by the character number (from 0 to 7), but immediately after that
these bytes are got rewritten by the decoded hex stream.

Fix the index to fill the buffer after the command and number.

Fixes: 8c9108d014c5 ("auxdisplay: add a driver for lcd2s character display")
Cc: Lars Poeschel <poeschel@lemonage.de>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
---
 drivers/auxdisplay/lcd2s.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/auxdisplay/lcd2s.c b/drivers/auxdisplay/lcd2s.c
index 38ba08628ccb..ea9d75ad4f16 100644
--- a/drivers/auxdisplay/lcd2s.c
+++ b/drivers/auxdisplay/lcd2s.c
@@ -238,7 +238,7 @@ static int lcd2s_redefine_char(struct charlcd *lcd, char *esc)
 	if (buf[1] > 7)
 		return 1;
 
-	i = 0;
+	i = 2;
 	shift = 0;
 	value = 0;
 	while (*esc && i < LCD2S_CHARACTER_SIZE + 2) {
-- 
2.34.1

[PATCH v1 2/3] auxdisplay: lcd2s: Fix memory leak in ->remove()

[Andy Shevchenko]

Once allocated the struct lcd2s_data is never freed.
Fix the memory leak by switching to devm_kzalloc().

Fixes: 8c9108d014c5 ("auxdisplay: add a driver for lcd2s character display")
Cc: Lars Poeschel <poeschel@lemonage.de>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
---
 drivers/auxdisplay/lcd2s.c | 18 +++++++-----------
 1 file changed, 7 insertions(+), 11 deletions(-)

diff --git a/drivers/auxdisplay/lcd2s.c b/drivers/auxdisplay/lcd2s.c
index ea9d75ad4f16..7b9cfbbaf007 100644
--- a/drivers/auxdisplay/lcd2s.c
+++ b/drivers/auxdisplay/lcd2s.c
@@ -298,6 +298,10 @@ static int lcd2s_i2c_probe(struct i2c_client *i2c,
 			I2C_FUNC_SMBUS_WRITE_BLOCK_DATA))
 		return -EIO;
 
+	lcd2s = devm_kzalloc(&i2c->dev, sizeof(*lcd2s), GFP_KERNEL);
+	if (!lcd2s)
+		return -ENOMEM;
+
 	/* Test, if the display is responding */
 	err = lcd2s_i2c_smbus_write_byte(i2c, LCD2S_CMD_DISPLAY_OFF);
 	if (err < 0)
@@ -307,12 +311,6 @@ static int lcd2s_i2c_probe(struct i2c_client *i2c,
 	if (!lcd)
 		return -ENOMEM;
 
-	lcd2s = kzalloc(sizeof(struct lcd2s_data), GFP_KERNEL);
-	if (!lcd2s) {
-		err = -ENOMEM;
-		goto fail1;
-	}
-
 	lcd->drvdata = lcd2s;
 	lcd2s->i2c = i2c;
 	lcd2s->charlcd = lcd;
@@ -321,24 +319,22 @@ static int lcd2s_i2c_probe(struct i2c_client *i2c,
 	err = device_property_read_u32(&i2c->dev, "display-height-chars",
 			&lcd->height);
 	if (err)
-		goto fail2;
+		goto fail1;
 
 	err = device_property_read_u32(&i2c->dev, "display-width-chars",
 			&lcd->width);
 	if (err)
-		goto fail2;
+		goto fail1;
 
 	lcd->ops = &lcd2s_ops;
 
 	err = charlcd_register(lcd2s->charlcd);
 	if (err)
-		goto fail2;
+		goto fail1;
 
 	i2c_set_clientdata(i2c, lcd2s);
 	return 0;
 
-fail2:
-	kfree(lcd2s);
 fail1:
 	kfree(lcd);
 	return err;
-- 
2.34.1

[PATCH v1 3/3] auxdisplay: lcd2s: Use proper API to free the instance of charlcd object

[Andy Shevchenko]

While it might work, the current approach is fragile in a few ways:
- whenever members in the structure are shuffled, the pointer will be wrong
- the resource freeing may include more than covered by kfree()

Fix this by using charlcd_free() call instead of kfree().

Fixes: 8c9108d014c5 ("auxdisplay: add a driver for lcd2s character display")
Cc: Lars Poeschel <poeschel@lemonage.de>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
---
 drivers/auxdisplay/lcd2s.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/auxdisplay/lcd2s.c b/drivers/auxdisplay/lcd2s.c
index 7b9cfbbaf007..2578b2d45439 100644
--- a/drivers/auxdisplay/lcd2s.c
+++ b/drivers/auxdisplay/lcd2s.c
@@ -336,7 +336,7 @@ static int lcd2s_i2c_probe(struct i2c_client *i2c,
 	return 0;
 
 fail1:
-	kfree(lcd);
+	charlcd_free(lcd2s->charlcd);
 	return err;
 }
 
@@ -345,7 +345,7 @@ static int lcd2s_i2c_remove(struct i2c_client *i2c)
 	struct lcd2s_data *lcd2s = i2c_get_clientdata(i2c);
 
 	charlcd_unregister(lcd2s->charlcd);
-	kfree(lcd2s->charlcd);
+	charlcd_free(lcd2s->charlcd);
 	return 0;
 }
 
-- 
2.34.1

Re: [PATCH v1 1/3] auxdisplay: lcd2s: Fix lcd2s_redefine_char() feature

[Andy Shevchenko]

On Wed, Feb 23, 2022 at 05:47:16PM +0200, Andy Shevchenko wrote:
> It seems that the lcd2s_redefine_char() has never been properly
> tested. The buffer is filled by DEF_CUSTOM_CHAR command followed
> by the character number (from 0 to 7), but immediately after that
> these bytes are got rewritten by the decoded hex stream.
> 
> Fix the index to fill the buffer after the command and number.

Any comments on this, please? There are fixes and seems auxdisplay subsystem
is a bit lagging... Should we worry about?

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH v1 1/3] auxdisplay: lcd2s: Fix lcd2s_redefine_char() feature

[Geert Uytterhoeven]

On Thu, Feb 24, 2022 at 2:03 AM Andy Shevchenko
<andriy.shevchenko@linux.intel.com> wrote:
> It seems that the lcd2s_redefine_char() has never been properly
> tested. The buffer is filled by DEF_CUSTOM_CHAR command followed
> by the character number (from 0 to 7), but immediately after that
> these bytes are got rewritten by the decoded hex stream.
>
> Fix the index to fill the buffer after the command and number.
>
> Fixes: 8c9108d014c5 ("auxdisplay: add a driver for lcd2s character display")
> Cc: Lars Poeschel <poeschel@lemonage.de>
> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>

Reviewed-by: Geert Uytterhoeven <geert@linux-m68k.org>

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: [PATCH v1 1/3] auxdisplay: lcd2s: Fix lcd2s_redefine_char() feature

[Miguel Ojeda]

On Wed, Feb 23, 2022 at 4:47 PM Andy Shevchenko
<andriy.shevchenko@linux.intel.com> wrote:
>
> It seems that the lcd2s_redefine_char() has never been properly
> tested. The buffer is filled by DEF_CUSTOM_CHAR command followed
> by the character number (from 0 to 7), but immediately after that
> these bytes are got rewritten by the decoded hex stream.
>
> Fix the index to fill the buffer after the command and number.

Thanks Andy, queued.

Cheers,
Miguel

Re: [PATCH v1 2/3] auxdisplay: lcd2s: Fix memory leak in ->remove()

[Miguel Ojeda]

On Wed, Feb 23, 2022 at 4:47 PM Andy Shevchenko
<andriy.shevchenko@linux.intel.com> wrote:
>
> Once allocated the struct lcd2s_data is never freed.
> Fix the memory leak by switching to devm_kzalloc().
>
> Fixes: 8c9108d014c5 ("auxdisplay: add a driver for lcd2s character display")
> Cc: Lars Poeschel <poeschel@lemonage.de>
> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>

Thanks Andy, queued.

Cheers,
Miguel

Re: [PATCH v1 3/3] auxdisplay: lcd2s: Use proper API to free the instance of charlcd object

[Miguel Ojeda]

On Wed, Feb 23, 2022 at 4:47 PM Andy Shevchenko
<andriy.shevchenko@linux.intel.com> wrote:
>
> While it might work, the current approach is fragile in a few ways:
> - whenever members in the structure are shuffled, the pointer will be wrong
> - the resource freeing may include more than covered by kfree()
>
> Fix this by using charlcd_free() call instead of kfree().
>
> Fixes: 8c9108d014c5 ("auxdisplay: add a driver for lcd2s character display")
> Cc: Lars Poeschel <poeschel@lemonage.de>
> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>

Thanks Andy, queued.

Cheers,
Miguel

Re: [PATCH v1 1/3] auxdisplay: lcd2s: Fix lcd2s_redefine_char() feature

[Miguel Ojeda]

On Thu, Mar 3, 2022 at 2:18 AM Miguel Ojeda
<miguel.ojeda.sandonis@gmail.com> wrote:
>
> Thanks Andy, queued.

By the way, I fixed a typo in the commit message.

Cheers,
Miguel

Re: [PATCH v1 1/3] auxdisplay: lcd2s: Fix lcd2s_redefine_char() feature

[Andy Shevchenko]

On Thu, Mar 03, 2022 at 11:46:32AM +0100, Miguel Ojeda wrote:
> On Thu, Mar 3, 2022 at 2:18 AM Miguel Ojeda
> <miguel.ojeda.sandonis@gmail.com> wrote:
> >
> > Thanks Andy, queued.
> 
> By the way, I fixed a typo in the commit message.

Thank you, Miguel!

-- 
With Best Regards,
Andy Shevchenko