[PATCH 1/2 4.19.y] ALSA: usb-audio: Sanity checks for each pipe and EP types
[ Upstream commit 801ebf1043ae7b182588554cc9b9ad3c14bc2ab5 ]
The recent USB core code performs sanity checks for the given pipe and EP types, and it can be hit by manipulated USB descriptors by syzbot. For making syzbot happier, this patch introduces a local helper for a sanity check in the driver side and calls it at each place before the message handling, so that we can avoid the WARNING splats.
Reported-by: syzbot+d952e5e28f5fb7718d23@syzkaller.appspotmail.com Signed-off-by: Takashi Iwai tiwai@suse.de Signed-off-by: Dan Carpenter dan.carpenter@linaro.org --- sound/usb/helper.c | 17 +++++++++++++++++ sound/usb/helper.h | 1 + sound/usb/quirks.c | 14 +++++++++++--- 3 files changed, 29 insertions(+), 3 deletions(-)
diff --git a/sound/usb/helper.c b/sound/usb/helper.c index 7712e2b84183..b1cc9499c57e 100644 --- a/sound/usb/helper.c +++ b/sound/usb/helper.c @@ -76,6 +76,20 @@ void *snd_usb_find_csint_desc(void *buffer, int buflen, void *after, u8 dsubtype return NULL; }
+/* check the validity of pipe and EP types */ +int snd_usb_pipe_sanity_check(struct usb_device *dev, unsigned int pipe) +{ + static const int pipetypes[4] = { + PIPE_CONTROL, PIPE_ISOCHRONOUS, PIPE_BULK, PIPE_INTERRUPT + }; + struct usb_host_endpoint *ep; + + ep = usb_pipe_endpoint(dev, pipe); + if (usb_pipetype(pipe) != pipetypes[usb_endpoint_type(&ep->desc)]) + return -EINVAL; + return 0; +} + /* * Wrapper for usb_control_msg(). * Allocates a temp buffer to prevent dmaing from/to the stack. @@ -88,6 +102,9 @@ int snd_usb_ctl_msg(struct usb_device *dev, unsigned int pipe, __u8 request, void *buf = NULL; int timeout;
+ if (snd_usb_pipe_sanity_check(dev, pipe)) + return -EINVAL; + if (size > 0) { buf = kmemdup(data, size, GFP_KERNEL); if (!buf) diff --git a/sound/usb/helper.h b/sound/usb/helper.h index f5b4c6647e4d..5e8a18b4e7b9 100644 --- a/sound/usb/helper.h +++ b/sound/usb/helper.h @@ -7,6 +7,7 @@ unsigned int snd_usb_combine_bytes(unsigned char *bytes, int size); void *snd_usb_find_desc(void *descstart, int desclen, void *after, u8 dtype); void *snd_usb_find_csint_desc(void *descstart, int desclen, void *after, u8 dsubtype);
+int snd_usb_pipe_sanity_check(struct usb_device *dev, unsigned int pipe); int snd_usb_ctl_msg(struct usb_device *dev, unsigned int pipe, __u8 request, __u8 requesttype, __u16 value, __u16 index, void *data, __u16 size); diff --git a/sound/usb/quirks.c b/sound/usb/quirks.c index 43cbaaff163f..087eef5e249d 100644 --- a/sound/usb/quirks.c +++ b/sound/usb/quirks.c @@ -743,11 +743,13 @@ static int snd_usb_novation_boot_quirk(struct usb_device *dev) static int snd_usb_accessmusic_boot_quirk(struct usb_device *dev) { int err, actual_length; - /* "midi send" enable */ static const u8 seq[] = { 0x4e, 0x73, 0x52, 0x01 }; + void *buf;
- void *buf = kmemdup(seq, ARRAY_SIZE(seq), GFP_KERNEL); + if (snd_usb_pipe_sanity_check(dev, usb_sndintpipe(dev, 0x05))) + return -EINVAL; + buf = kmemdup(seq, ARRAY_SIZE(seq), GFP_KERNEL); if (!buf) return -ENOMEM; err = usb_interrupt_msg(dev, usb_sndintpipe(dev, 0x05), buf, @@ -772,7 +774,11 @@ static int snd_usb_accessmusic_boot_quirk(struct usb_device *dev)
static int snd_usb_nativeinstruments_boot_quirk(struct usb_device *dev) { - int ret = usb_control_msg(dev, usb_sndctrlpipe(dev, 0), + int ret; + + if (snd_usb_pipe_sanity_check(dev, usb_sndctrlpipe(dev, 0))) + return -EINVAL; + ret = usb_control_msg(dev, usb_sndctrlpipe(dev, 0), 0xaf, USB_TYPE_VENDOR | USB_RECIP_DEVICE, 1, 0, NULL, 0, 1000);
@@ -879,6 +885,8 @@ static int snd_usb_axefx3_boot_quirk(struct usb_device *dev)
dev_dbg(&dev->dev, "Waiting for Axe-Fx III to boot up...\n");
+ if (snd_usb_pipe_sanity_check(dev, usb_sndctrlpipe(dev, 0))) + return -EINVAL; /* If the Axe-Fx III has not fully booted, it will timeout when trying * to enable the audio streaming interface. A more generous timeout is * used here to detect when the Axe-Fx III has finished booting as the
This is a note to let you know that I've just added the patch titled
ALSA: usb-audio: Sanity checks for each pipe and EP types
to the 4.19-stable tree which can be found at: http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summ...
The filename of the patch is: alsa-usb-audio-sanity-checks-for-each-pipe-and-ep-types.patch and it can be found in the queue-4.19 subdirectory.
If you, or anyone else, feels it should not be added to the stable tree, please let stable@vger.kernel.org know about it.
From stable+bounces-73596-greg=kroah.com@vger.kernel.org Thu Sep 5 14:38:22 2024 From: Takashi Iwai dan.carpenter@linaro.org Date: Thu, 5 Sep 2024 15:38:07 +0300 Subject: ALSA: usb-audio: Sanity checks for each pipe and EP types To: Jaroslav Kysela perex@perex.cz Cc: Takashi Iwai tiwai@suse.com, Hillf Danton hdanton@sina.com, alsa-devel@alsa-project.org, stable@vger.kernel.org Message-ID: 599b79d0-0c0f-425e-b2a2-1af9f81539b8@stanley.mountain Content-Disposition: inline
From: Takashi Iwai dan.carpenter@linaro.org
[ Upstream commit 801ebf1043ae7b182588554cc9b9ad3c14bc2ab5 ]
The recent USB core code performs sanity checks for the given pipe and EP types, and it can be hit by manipulated USB descriptors by syzbot. For making syzbot happier, this patch introduces a local helper for a sanity check in the driver side and calls it at each place before the message handling, so that we can avoid the WARNING splats.
Reported-by: syzbot+d952e5e28f5fb7718d23@syzkaller.appspotmail.com Signed-off-by: Takashi Iwai tiwai@suse.de Signed-off-by: Dan Carpenter dan.carpenter@linaro.org Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- sound/usb/helper.c | 17 +++++++++++++++++ sound/usb/helper.h | 1 + sound/usb/quirks.c | 14 +++++++++++--- 3 files changed, 29 insertions(+), 3 deletions(-)
--- a/sound/usb/helper.c +++ b/sound/usb/helper.c @@ -76,6 +76,20 @@ void *snd_usb_find_csint_desc(void *buff return NULL; }
+/* check the validity of pipe and EP types */ +int snd_usb_pipe_sanity_check(struct usb_device *dev, unsigned int pipe) +{ + static const int pipetypes[4] = { + PIPE_CONTROL, PIPE_ISOCHRONOUS, PIPE_BULK, PIPE_INTERRUPT + }; + struct usb_host_endpoint *ep; + + ep = usb_pipe_endpoint(dev, pipe); + if (usb_pipetype(pipe) != pipetypes[usb_endpoint_type(&ep->desc)]) + return -EINVAL; + return 0; +} + /* * Wrapper for usb_control_msg(). * Allocates a temp buffer to prevent dmaing from/to the stack. @@ -88,6 +102,9 @@ int snd_usb_ctl_msg(struct usb_device *d void *buf = NULL; int timeout;
+ if (snd_usb_pipe_sanity_check(dev, pipe)) + return -EINVAL; + if (size > 0) { buf = kmemdup(data, size, GFP_KERNEL); if (!buf) --- a/sound/usb/helper.h +++ b/sound/usb/helper.h @@ -7,6 +7,7 @@ unsigned int snd_usb_combine_bytes(unsig void *snd_usb_find_desc(void *descstart, int desclen, void *after, u8 dtype); void *snd_usb_find_csint_desc(void *descstart, int desclen, void *after, u8 dsubtype);
+int snd_usb_pipe_sanity_check(struct usb_device *dev, unsigned int pipe); int snd_usb_ctl_msg(struct usb_device *dev, unsigned int pipe, __u8 request, __u8 requesttype, __u16 value, __u16 index, void *data, __u16 size); --- a/sound/usb/quirks.c +++ b/sound/usb/quirks.c @@ -743,11 +743,13 @@ static int snd_usb_novation_boot_quirk(s static int snd_usb_accessmusic_boot_quirk(struct usb_device *dev) { int err, actual_length; - /* "midi send" enable */ static const u8 seq[] = { 0x4e, 0x73, 0x52, 0x01 }; + void *buf;
- void *buf = kmemdup(seq, ARRAY_SIZE(seq), GFP_KERNEL); + if (snd_usb_pipe_sanity_check(dev, usb_sndintpipe(dev, 0x05))) + return -EINVAL; + buf = kmemdup(seq, ARRAY_SIZE(seq), GFP_KERNEL); if (!buf) return -ENOMEM; err = usb_interrupt_msg(dev, usb_sndintpipe(dev, 0x05), buf, @@ -772,7 +774,11 @@ static int snd_usb_accessmusic_boot_quir
static int snd_usb_nativeinstruments_boot_quirk(struct usb_device *dev) { - int ret = usb_control_msg(dev, usb_sndctrlpipe(dev, 0), + int ret; + + if (snd_usb_pipe_sanity_check(dev, usb_sndctrlpipe(dev, 0))) + return -EINVAL; + ret = usb_control_msg(dev, usb_sndctrlpipe(dev, 0), 0xaf, USB_TYPE_VENDOR | USB_RECIP_DEVICE, 1, 0, NULL, 0, 1000);
@@ -879,6 +885,8 @@ static int snd_usb_axefx3_boot_quirk(str
dev_dbg(&dev->dev, "Waiting for Axe-Fx III to boot up...\n");
+ if (snd_usb_pipe_sanity_check(dev, usb_sndctrlpipe(dev, 0))) + return -EINVAL; /* If the Axe-Fx III has not fully booted, it will timeout when trying * to enable the audio streaming interface. A more generous timeout is * used here to detect when the Axe-Fx III has finished booting as the
Patches currently in stable-queue which might be from dan.carpenter@linaro.org are
queue-4.19/alsa-usb-audio-sanity-checks-for-each-pipe-and-ep-types.patch queue-4.19/alsa-usb-audio-fix-gpf-in-snd_usb_pipe_sanity_check.patch
Sorry,
I completely messed these emails up. It has Takashi Iwai and Hillf Danton's names instead of mine in the From header. It still has my email address, but just the names are wrong.
Also I should have used a From header in the body of the email.
Also the threading is messed up.
Will try again tomorrow.
regards, dan carpenter
On Thu, Sep 05, 2024 at 04:34:45PM +0300, Dan Carpenter wrote:
Sorry,
I completely messed these emails up. It has Takashi Iwai and Hillf Danton's names instead of mine in the From header. It still has my email address, but just the names are wrong.
Also I should have used a From header in the body of the email.
Also the threading is messed up.
Will try again tomorrow.
It looks good to me, now queued up.
thanks,
greg k-h
On Thu, Sep 05, 2024 at 03:49:14PM +0200, Greg KH wrote:
On Thu, Sep 05, 2024 at 04:34:45PM +0300, Dan Carpenter wrote:
Sorry,
I completely messed these emails up. It has Takashi Iwai and Hillf Danton's names instead of mine in the From header. It still has my email address, but just the names are wrong.
Also I should have used a From header in the body of the email.
Also the threading is messed up.
Will try again tomorrow.
It looks good to me, now queued up.
The code is okay but the Author header is messed up. It has my email address.
From: Hillf Danton dan.carpenter@linaro.org
From: Takashi Iwai dan.carpenter@linaro.org
regards, dan carpenter
On Thu, Sep 05, 2024 at 06:11:49PM +0300, Dan Carpenter wrote:
On Thu, Sep 05, 2024 at 03:49:14PM +0200, Greg KH wrote:
On Thu, Sep 05, 2024 at 04:34:45PM +0300, Dan Carpenter wrote:
Sorry,
I completely messed these emails up. It has Takashi Iwai and Hillf Danton's names instead of mine in the From header. It still has my email address, but just the names are wrong.
Also I should have used a From header in the body of the email.
Also the threading is messed up.
Will try again tomorrow.
It looks good to me, now queued up.
The code is okay but the Author header is messed up. It has my email address.
From: Hillf Danton dan.carpenter@linaro.org
From: Takashi Iwai dan.carpenter@linaro.org
Ah that's really odd, how did you do that? :)
Now fixed up in the patches, don't worry about it.
greg k-h
participants (4)
-
Dan Carpenter -
Greg KH -
gregkh＠linuxfoundation.org
-
Takashi Iwai