HID: roccat: add bounds checking in kone_sysfs_write_settings()
Subsystems (3)
PATCH 1/1 HID: roccat: add bounds checking in kone_sysfs_write_settings()
In the original code we didn't check if the new value for "settings->startup_profile" was within bounds that could possibly result in an out of bounds array acccess. What we did was we checked if we could write the data to the firmware and it's possibly the firmware checks these values but there is no way to know. It's safer and easier to read if we check it in the kernel as well.
I also added a check to ensure that "settings->size" was correct. The comments say that the only valid value is 36 which is sizeof(struct kone_settings).
---
This isn't tested. Checking settings->size seems like a good idea, but there is a slight risky because maybe invalid values used to be allowed and I don't want to break user space.
drivers/hid/hid-roccat-kone.c | 24 +++++++++++++++++------- 1 file changed, 17 insertions(+), 7 deletions(-)
drivers/hid/hid-roccat-kone.c +17 -7
@@ -294,31 +294,41 @@ static ssize_t kone_sysfs_write_settings(struct file *fp, struct kobject *kobj,struct kone_device *kone =hid_get_drvdata (dev_get_drvdata (dev));struct usb_device *usb_dev =interface_to_usbdev (to_usb_interface (dev));int retval =0 , difference, old_profile; +struct kone_settings *settings = (struct kone_settings *)buf;/* I need to get my data in one piece */ if (off !=0 || count !=sizeof (struct kone_settings ))return -EINVAL ;mutex_lock (&kone->kone_lock); - difference =memcmp (buf, &kone->settings,sizeof (struct kone_settings )); + difference =memcmp (settings, &kone->settings, +sizeof (struct kone_settings ));if (difference) { - retval =kone_set_settings (usb_dev, - (struct kone_settings const *)buf); -if (retval) { -mutex_unlock (&kone->kone_lock); -return retval; +if (settings->size !=sizeof (struct kone_settings ) || + settings->startup_profile <1 || + settings->startup_profile >5 ) { + retval = -EINVAL ; + goto unlock; } + retval =kone_set_settings (usb_dev, settings); +if (retval) + goto unlock; + old_profile = kone->settings.startup_profile; -memcpy (&kone->settings, buf,sizeof (struct kone_settings )); +memcpy (&kone->settings, settings,sizeof (struct kone_settings ));kone_profile_activated (kone, kone->settings.startup_profile);if (kone->settings.startup_profile != old_profile)kone_profile_report (kone, kone->settings.startup_profile); } +unlock:mutex_unlock (&kone->kone_lock); +if (retval) +return retval; +return sizeof (struct kone_settings ); } static BIN_ATTR(settings, 0660, kone_sysfs_read_settings,
Signature · 1 line
2.27.0
Jiri Kosina
maintainerQuoted 7 lines · Dan Carpenter
In the original code we didn't check if the new value for "
settings->startup_profile" was within bounds that could possibly result in an out of bounds array acccess. What we did was we checked if we could write the data to the firmware and it's possibly the firmware checks these values but there is no way to know. It's safer and easier to read if we check it in the kernel as well.
Quoted 6 lines
I also added a check to ensure that "
settings->size" was correct. The comments say that the only valid value is 36 which issizeof(struct kone_settings).Fixes 14bf62cde794 ("HID: add driver for Roccat Kone gaming mouse")Signed-off-by Dan Carpenter <[email protected]>
Stefan, could I please get your Reviewed-by and/or Tested-by, to make sure this doesn't unintentionally somehow break userspace?
Thanks,
Signature · 2 lines
Jiri Kosina SUSE Labs
Please remove the settings->size check. The driver sometimes just writes a modified version of the read settings data. This would fail if stored size is invalid. This value of size is not solely in my hands, I can't guarantee Windows driver does a sanity check, also cases of flash data corruption are known.
My general design agenda is that in my userspace tools I make sure only valid data (to the best of knowledge) is prepared for the device. In the kernel driver I don't (or didn't) do additional checks and eventually let the hardware reject things it doesn't like. This way the kernel driver doesn't need an update if firmware updates change the binary interface. This happend in the past, but won't for this device anymore because it's oop for years.
Am Mittwoch, den 05.08.2020, 12:55 +0300 schrieb Dan Carpenter:
Quoted 91 lines
In the original code we didn't check if the new value for "
settings->startup_profile" was within bounds that could possibly result in an out of bounds array acccess. What we did was we checked if
we could write the data to the firmware and it's possibly the firmware
checks these values but there is no way to know. It's safer and easier
to read if we check it in the kernel as well.I also added a check to ensure that "
settings->size" was correct. The
comments say that the only valid value is 36 which issizeof(struct kone_settings).Fixes 14bf62cde794 ("HID: add driver for Roccat Kone gaming mouse")Signed-off-by Dan Carpenter <[email protected]>---
This isn't tested. Checkingsettings->sizeseems like a good idea, but
there is a slight risky because maybe invalid values used to be allowed
and I don't want to break user space.drivers/hid/hid-roccat-kone.c | 24 +++++++++++++++++------- 1 file changed, 17 insertions(+), 7 deletions(-)drivers/hid/hid-roccat- renamed from drivers/hid/hid-roccat-kone.c +0
-0
kone.c
index1a6e600197d0..5e8e1517f23c100644drivers/hid/hid-roccat-kone.c +4
-6@@ -294,31 +294,41 @@ static ssize_t kone_sysfs_write_settings(struct file *fp, struct kobject *kobj, struct kone_device *kone =hid_get_drvdata (dev_get_drvdata (dev));struct usb_device *usb_dev =interface_to_usbdev (to_usb_interface (dev));int retval =0 , difference, old_profile; +struct kone_settings *settings = (struct kone_settings *)buf;/* I need to get my data in one piece */ if (off !=0 || count !=sizeof (struct kone_settings ))return -EINVAL ;mutex_lock (&kone->kone_lock); - difference =memcmp (buf, &kone->settings,sizeof (struct kone_settings)); + difference =memcmp (settings, &kone->settings, +sizeof (struct kone_settings ));if (difference) { - retval =kone_set_settings (usb_dev, - (struct kone_settings const *)buf); -if (retval) { -mutex_unlock (&kone->kone_lock); -return retval; + if (settings->size != sizeof(struct kone_settings)||
drivers/hid/hid-roccat-kone.c +8
-1+ settings->startup_profile < 1 || + settings->startup_profile > 5) { + retval = -EINVAL ; + goto unlock; } + retval =kone_set_settings (usb_dev, settings); +if (retval) + goto unlock; + old_profile = kone->settings.startup_profile; - memcpy(&kone->settings, buf, sizeof(struct
kone_settings));drivers/hid/hid-roccat-kone.c +1
-0+ memcpy(&kone->settings, settings, sizeof(struct
kone_settings));kone_profile_activated(kone, kone-
Quoted 1 line
settings.startup_profile);
drivers/hid/hid-roccat-kone.c +0
-0if (kone->settings.startup_profile != old_profile) kone_profile_report(kone, kone-Quoted 1 line
settings.startup_profile);
drivers/hid/hid-roccat-kone.c +4
-0} +unlock:mutex_unlock (&kone->kone_lock); +if (retval) +return retval; +return sizeof (struct kone_settings ); } static BIN_ATTR(settings, 0660, kone_sysfs_read_settings,
PATCH v2 1/1 HID: roccat: add bounds checking in kone_sysfs_write_settings()
This code doesn't check if "settings->startup_profile" is within bounds and that could result in an out of bounds array access. What the code does do is it checks if the settings can be written to the firmware, so it's possible that the firmware has a bounds check? It's safer and easier to verify when the bounds checking is done in the kernel.
---
v2: In the v1 patch I added a check against settings->size but that's potentially too strict so it was removed.
drivers/hid/hid-roccat-kone.c | 23 ++++++++++++++++------- 1 file changed, 16 insertions(+), 7 deletions(-)
drivers/hid/hid-roccat-kone.c +16 -7
@@ -294,31 +294,40 @@ static ssize_t kone_sysfs_write_settings(struct file *fp, struct kobject *kobj,struct kone_device *kone =hid_get_drvdata (dev_get_drvdata (dev));struct usb_device *usb_dev =interface_to_usbdev (to_usb_interface (dev));int retval =0 , difference, old_profile; +struct kone_settings *settings = (struct kone_settings *)buf;/* I need to get my data in one piece */ if (off !=0 || count !=sizeof (struct kone_settings ))return -EINVAL ;mutex_lock (&kone->kone_lock); - difference =memcmp (buf, &kone->settings,sizeof (struct kone_settings )); + difference =memcmp (settings, &kone->settings, +sizeof (struct kone_settings ));if (difference) { - retval =kone_set_settings (usb_dev, - (struct kone_settings const *)buf); -if (retval) { -mutex_unlock (&kone->kone_lock); -return retval; +if (settings->startup_profile <1 || + settings->startup_profile >5 ) { + retval = -EINVAL ; + goto unlock; } + retval =kone_set_settings (usb_dev, settings); +if (retval) + goto unlock; + old_profile = kone->settings.startup_profile; -memcpy (&kone->settings, buf,sizeof (struct kone_settings )); +memcpy (&kone->settings, settings,sizeof (struct kone_settings ));kone_profile_activated (kone, kone->settings.startup_profile);if (kone->settings.startup_profile != old_profile)kone_profile_report (kone, kone->settings.startup_profile); } +unlock:mutex_unlock (&kone->kone_lock); +if (retval) +return retval; +return sizeof (struct kone_settings ); } static BIN_ATTR(settings, 0660, kone_sysfs_read_settings,
Signature · 1 line
2.28.0
hello Dan,
i notice that you can shorten the line to:
(line above checks for count==sizeof(struct kone_settings))
difference = memcmp(settings, &kone->settings, count);
nothing special just to shorten the line and make use of count.
and just to save one indent level and because its readabel nicely:
if ( ! difference )
goto unlock;
hope that helps
re,
wh
________________________________________
Von: [email protected] [[email protected]] im Auftrag von Dan Carpenter [[email protected]]
Gesendet: Montag, 24. August 2020 10:57
An: Stefan Achatz
Cc: Jiri Kosina; Benjamin Tissoires; [email protected]; [email protected]; [email protected]
Betreff: [PATCH v2] HID: roccat: add bounds checking in kone_sysfs_write_settings()
This code doesn't check if "settings->startup_profile" is within bounds and that could result in an out of bounds array access. What the code does do is it checks if the settings can be written to the firmware, so it's possible that the firmware has a bounds check? It's safer and easier to verify when the bounds checking is done in the kernel.
---
v2: In the v1 patch I added a check against settings->size but that's potentially too strict so it was removed.
drivers/hid/hid-roccat-kone.c | 23 ++++++++++++++++------- 1 file changed, 16 insertions(+), 7 deletions(-)
drivers/hid/hid-roccat-kone.c +16 -7
@@ -294,31 +294,40 @@ static ssize_t kone_sysfs_write_settings(struct file *fp, struct kobject *kobj,struct kone_device *kone =hid_get_drvdata (dev_get_drvdata (dev));struct usb_device *usb_dev =interface_to_usbdev (to_usb_interface (dev));int retval =0 , difference, old_profile; +struct kone_settings *settings = (struct kone_settings *)buf;/* I need to get my data in one piece */ if (off !=0 || count !=sizeof (struct kone_settings ))return -EINVAL ;mutex_lock (&kone->kone_lock); - difference =memcmp (buf, &kone->settings,sizeof (struct kone_settings )); + difference =memcmp (settings, &kone->settings, +sizeof (struct kone_settings ));if (difference) { - retval =kone_set_settings (usb_dev, - (struct kone_settings const *)buf); -if (retval) { -mutex_unlock (&kone->kone_lock); -return retval; +if (settings->startup_profile <1 || + settings->startup_profile >5 ) { + retval = -EINVAL ; + goto unlock; } + retval =kone_set_settings (usb_dev, settings); +if (retval) + goto unlock; + old_profile = kone->settings.startup_profile; -memcpy (&kone->settings, buf,sizeof (struct kone_settings )); +memcpy (&kone->settings, settings,sizeof (struct kone_settings ));kone_profile_activated (kone, kone->settings.startup_profile);if (kone->settings.startup_profile != old_profile)kone_profile_report (kone, kone->settings.startup_profile); } +unlock:mutex_unlock (&kone->kone_lock); +if (retval) +return retval; +return sizeof (struct kone_settings ); } static BIN_ATTR(settings, 0660, kone_sysfs_read_settings,
--
2.28.0
Quoted 7 lines · Walter Harms
hello Dan,
i notice that you can shorten the line to:
(line above checks for count=sizeof(structkone_settings))difference =
memcmp(settings, &kone->settings, count);
Quoted 7 lines
nothing special just to shorten the line and make use of count.
and just to save one indent level and because its readabel nicely:
if ( ! difference )
goto unlock;hope that helps
Yeah. I wrote that version and I wanted to send it, but then I decided not to change the style too much. I definitely agree with you, but I figured I would keep the patch less intrusive.
regards,
dan carpenter
lets hope the maintainer picks that up.
re,
wh
________________________________________
Von: Dan Carpenter [[email protected]]
Gesendet: Dienstag, 25. August 2020 09:29
An: Walter Harms
Cc: Stefan Achatz; Jiri Kosina; Benjamin Tissoires; [email protected]; [email protected]; [email protected]
Betreff: Re: [PATCH v2] HID: roccat: add bounds checking in kone_sysfs_write_settings()
Quoted 7 lines · Walter Harms
hello Dan,
i notice that you can shorten the line to:
(line above checks for count==sizeof(structkone_settings))difference =
memcmp(settings, &kone->settings, count);
Quoted 7 lines
nothing special just to shorten the line and make use of count.
and just to save one indent level and because its readabel nicely:
if ( ! difference )
goto unlock;hope that helps
Yeah. I wrote that version and I wanted to send it, but then I decided not to change the style too much. I definitely agree with you, but I figured I would keep the patch less intrusive.
regards,
dan carpenter
Quoted 11 lines · Dan Carpenter
This code doesn't check if "
settings->startup_profile" is within bounds and that could result in an out of bounds array access. What the code does do is it checks if the settings can be written to the firmware, so it's possible that the firmware has a bounds check? It's safer and easier to verify when the bounds checking is done in the kernel.Fixes 14bf62cde794 ("HID: add driver for Roccat Kone gaming mouse")Signed-off-by Dan Carpenter <[email protected]>---
v2: In the v1 patch I added a check againstsettings->sizebut that's potentially too strict so it was removed.
Applied, thanks Dan.
Signature · 2 lines
Jiri Kosina SUSE Labs