ALSA: Don't embed struct devices
cover ALSA: Don't embed struct devices
Hi,
this is another set of patches to attempt papering over the UAF problems that are seen when the delayed kobject release is enabled, as initially reported by Curtis:
https://lore.kernel.org/r/[email protected]
There was a previous patch set with a different approach (using the device refcount dependencies), but this is a sort of step-back to the old way.
https://lore.kernel.org/r/[email protected]
After discussions and evaluations, we agreed that decoupling the struct device from each sound component object is the safest (and easiest) way as of now. For applying the changes more consistently, I introduced a new helper for the struct device allocation and initialization, and applied all components.
A couple of more changes for card_dev refcount managed aren't included in this patch set, though. They might be good to have, but this patch set should suffice for the currently seen UAF problems.
For a long-term solution, we may restructure the device management, then the struct devices may be embedded again in each object. But, it'll need lots of other changes and cleanups, a big TODO.
The latest patches are found in topic/dev-split branch of sound.git tree.
Takashi
===
Takashi Iwai (9):
ALSA: core: Introduce snd_device_alloc()
ALSA: control: Don't embed ctl_dev
ALSA: pcm: Don't embed device
ALSA: hwdep: Don't embed device
ALSA: rawmidi: Don't embed device
ALSA: compress: Don't embed device
ALSA: timer: Create device with snd_device_alloc()
ALSA: seq: Create device with snd_device_alloc()
ALSA: core: Drop snd_device_initialize()
include/sound/compress_driver.h | 2 +- include/sound/core.h | 4 ++-- include/sound/hwdep.h | 2 +- include/sound/pcm.h | 2 +- include/sound/rawmidi.h | 2 +- sound/aoa/soundbus/i2sbus/pcm.c | 4 ++-- sound/core/compress_offload.c | 16 ++++++++------ sound/core/control.c | 14 ++++++------ sound/core/control_led.c | 4 ++-- sound/core/hwdep.c | 38 ++++++++++++++++++--------------- sound/core/init.c | 28 +++++++++++++++--------- sound/core/pcm.c | 22 +++++++++++-------- sound/core/rawmidi.c | 29 +++++++++++-------------- sound/core/seq/seq_clientmgr.c | 16 ++++++++------ sound/core/timer.c | 16 ++++++++------ sound/core/ump.c | 8 +++---- sound/pci/hda/hda_hwdep.c | 4 ++-- sound/usb/media.c | 4 ++-- 18 files changed, 119 insertions(+), 96 deletions(-)
Signature · 1 line
2.35.3
RFC PATCH 1/9 ALSA: core: Introduce snd_device_alloc()
Introduce a new helper, snd_device_alloc(), for allocating a struct device that is bound with the sound class. It's a replacement of snd_device_initialize().
--- include/sound/core.h | 1 + sound/core/init.c | 31 +++++++++++++++++++++++++++++++ 2 files changed, 32 insertions(+)
include/sound/core.h +1 -0
@@ -239,6 +239,7 @@ extern struct dentry *sound_debugfs_root;void snd_request_card (int card); +int snd_device_alloc (struct device **dev_p,struct snd_card *card);void snd_device_initialize (struct device *dev,struct snd_card *card); int snd_register_device(int type, struct snd_card *card, int dev,
sound/core/init.c +31 -0
@@ -134,6 +134,37 @@ void snd_device_initialize(struct device *dev, struct snd_card *card) }EXPORT_SYMBOL_GPL (snd_device_initialize); +/* the default release callback set in snd_device_alloc() */ +static void default_release_alloc (struct device *dev) +{ +kfree (dev); +} + +/** +* snd_device_alloc - Allocate and initialize struct device for sound devices +* @dev_p: pointer to store the allocated device +* @card: card to assign, optional +* +* For releasing the allocated device, call put_device(). +*/ +int snd_device_alloc (struct device **dev_p,struct snd_card *card) +{ +struct device *dev; + + *dev_p =NULL ; + dev =kzalloc (sizeof (*dev),GFP_KERNEL ); +if (!dev) +return -ENOMEM ; +device_initialize (dev); +if (card) + dev->parent = &card->card_dev; + dev->class = &sound_class; + dev->release = default_release_alloc; + *dev_p = dev; +return 0 ; +} +EXPORT_SYMBOL_GPL (snd_device_alloc); +static int snd_card_init (struct snd_card *card,struct device *parent,int idx,const char *xid,struct module *module,size_t extra_size);
Signature · 1 line
2.35.3
RFC PATCH 2/9 ALSA: control: Don't embed ctl_dev
In reply to Takashi IwaiEmbedding the ctl_dev in the snd_card object may result in UAF when the delayed kobj release is used; at the delayed kobj release, it still accesses the struct device itself while the card memory (that embeds the struct device) may be already gone.
As a workaround, detach the struct device from the card object by allocating via the new snd_device_alloc() helper. The rest are just replacing ctl_dev access to the pointer.
This is based on the fix Curtis posted initially. In this patch, the changes are split and use the new helper function instead.
--- include/sound/core.h | 2 +- sound/core/control.c | 14 ++++++++------ sound/core/control_led.c | 4 ++-- sound/usb/media.c | 2 +- 4 files changed, 12 insertions(+), 10 deletions(-)
include/sound/core.h +1 -1
@@ -96,7 +96,7 @@ struct snd_card {private data */ struct list_head devices;/* devices */ -struct device ctl_dev;/* control device */ +struct device *ctl_dev;/* control device */ unsigned int last_numid;/* last used numeric ID */ struct rw_semaphore controls_rwsem;/* controls lock (list and values) */ rwlock_t ctl_files_rwlock;/* ctl_files list lock */
sound/core/control.c +8 -6
@@ -2315,7 +2315,7 @@ static int snd_ctl_dev_register(struct snd_device *device)int err; err =snd_register_device (SNDRV_DEVICE_TYPE_CONTROL , card,-1 , - &snd_ctl_f_ops, card, &card->ctl_dev); + &snd_ctl_f_ops, card, card->ctl_dev);if (err <0 )return err;down_read (&card->controls_rwsem);
@@ -2351,7 +2351,7 @@ static int snd_ctl_dev_disconnect(struct snd_device *device)up_read (&snd_ctl_layer_rwsem);up_read (&card->controls_rwsem); -return snd_unregister_device (&card->ctl_dev); +return snd_unregister_device (card->ctl_dev); }/*
@@ -2373,7 +2373,7 @@ static int snd_ctl_dev_free(struct snd_device *device)xa_destroy (&card->ctl_hash);#endif up_write (&card->controls_rwsem); -put_device (&card->ctl_dev); +put_device (card->ctl_dev);return 0 ; }
@@ -2395,12 +2395,14 @@ int snd_ctl_create(struct snd_card *card)if (snd_BUG_ON (card->number <0 || card->number >=SNDRV_CARDS ))return -ENXIO ; -snd_device_initialize (&card->ctl_dev, card); -dev_set_name (&card->ctl_dev,"controlC%d" , card->number); + err =snd_device_alloc (&card->ctl_dev, card); +if (err <0 ) +return err; +dev_set_name (card->ctl_dev,"controlC%d" , card->number); err =snd_device_new (card,SNDRV_DEV_CONTROL , card, &ops);if (err <0 ) -put_device (&card->ctl_dev); +put_device (card->ctl_dev);return err; }
sound/core/control_led.c +2 -2
@@ -688,7 +688,7 @@ static void snd_ctl_led_sysfs_add(struct snd_card *card) goto cerr; led->cards[card->number] = led_card;snprintf (link_name,sizeof (link_name),"led-%s" , led->name); -WARN (sysfs_create_link (&card->ctl_dev.kobj, &led_card->dev.kobj, link_name), +WARN (sysfs_create_link (&card->ctl_dev->kobj, &led_card->dev.kobj, link_name),"can't create symlink to controlC%i device\n" , card->number);WARN (sysfs_create_link (&led_card->dev.kobj, &card->card_dev.kobj,"card" ),"can't create symlink to card%i\n" , card->number);
@@ -714,7 +714,7 @@ static void snd_ctl_led_sysfs_remove(struct snd_card *card)if (!led_card)continue ;snprintf (link_name,sizeof (link_name),"led-%s" , led->name); -sysfs_remove_link (&card->ctl_dev.kobj, link_name); +sysfs_remove_link (&card->ctl_dev->kobj, link_name);sysfs_remove_link (&led_card->dev.kobj,"card" );device_unregister (&led_card->dev); led->cards[card->number] =NULL ;
sound/usb/media.c +1 -1
@@ -163,7 +163,7 @@ void snd_media_stop_pipeline(struct snd_usb_substream *subs)static int snd_media_mixer_init (struct snd_usb_audio *chip) { -struct device *ctl_dev = &chip->card->ctl_dev; +struct device *ctl_dev = chip->card->ctl_dev;struct media_intf_devnode *ctl_intf;struct usb_mixer_interface *mixer;struct media_device *mdev = chip->media_dev;
Signature · 1 line
2.35.3
RFC PATCH 3/9 ALSA: pcm: Don't embed device
In reply to Takashi IwaiSo far we use the embedded struct device for each PCM substreams in struct snd_pcm. This may result in UAF when the delayed kobj release is used; each corresponding struct device is still accessed at the (delayed) device release, while the snd_pcm object may be already gone.
As a workaround, detach the struct device from the snd_pcm object by allocating via the new snd_device_alloc() helper.
A caveat is that we store the PCM substream pointer to drvdata since the device resume and others require the access to it.
This patch is based on the fix Curtis posted initially. In this patch, the changes are split and use the new helper function instead.
--- include/sound/pcm.h | 2 +- sound/aoa/soundbus/i2sbus/pcm.c | 4 ++-- sound/core/pcm.c | 22 +++++++++++++--------- sound/usb/media.c | 2 +- 4 files changed, 17 insertions(+), 13 deletions(-)
include/sound/pcm.h +1 -1
@@ -510,7 +510,7 @@ struct snd_pcm_str {#endif #endif struct snd_kcontrol *chmap_kctl;/* channel-mapping controls */ -struct device dev; +struct device *dev; };struct snd_pcm {
sound/aoa/soundbus/i2sbus/pcm.c +2 -2
@@ -972,7 +972,7 @@ i2sbus_attach_codec(struct soundbus_dev *dev, struct snd_card *card, goto out_put_ci_module;snd_pcm_set_ops (dev->pcm,SNDRV_PCM_STREAM_PLAYBACK , &i2sbus_playback_ops); - dev->pcm->streams[SNDRV_PCM_STREAM_PLAYBACK ].dev.parent = + dev->pcm->streams[SNDRV_PCM_STREAM_PLAYBACK ]->dev.parent = &dev->ofdev.dev; i2sdev->out.created =1 ; }
@@ -989,7 +989,7 @@ i2sbus_attach_codec(struct soundbus_dev *dev, struct snd_card *card, goto out_put_ci_module;snd_pcm_set_ops (dev->pcm,SNDRV_PCM_STREAM_CAPTURE , &i2sbus_record_ops); - dev->pcm->streams[SNDRV_PCM_STREAM_CAPTURE ].dev.parent = + dev->pcm->streams[SNDRV_PCM_STREAM_CAPTURE ]->dev.parent = &dev->ofdev.dev; i2sdev->in.created =1 ; }
sound/core/pcm.c +13 -9
@@ -604,7 +604,7 @@ static const struct attribute_group *pcm_dev_attr_groups[];#ifdef CONFIG_PM_SLEEP static int do_pcm_suspend (struct device *dev) { - struct snd_pcm_str *pstr = container_of(dev, struct snd_pcm_str, dev); +struct snd_pcm_str *pstr =dev_get_drvdata (dev);if (!pstr->pcm->no_device_suspend)snd_pcm_suspend_all (pstr->pcm);
@@ -650,11 +650,14 @@ int snd_pcm_new_stream(struct snd_pcm *pcm, int stream, int substream_count)if (!substream_count)return 0 ; -snd_device_initialize (&pstr->dev, pcm->card); - pstr->dev.groups = pcm_dev_attr_groups; - pstr->dev.type = &pcm_dev_type; -dev_set_name (&pstr->dev,"pcmC%iD%i%c" , pcm->card->number, pcm->device, + err =snd_device_alloc (&pstr->dev, pcm->card); +if (err <0 ) +return err; +dev_set_name (pstr->dev,"pcmC%iD%i%c" , pcm->card->number, pcm->device, stream ==SNDRV_PCM_STREAM_PLAYBACK ?'p' :'c' ); + pstr->dev->groups = pcm_dev_attr_groups; + pstr->dev->type = &pcm_dev_type; +dev_set_drvdata (pstr->dev, pstr);if (!pcm->internal) { err =snd_pcm_stream_proc_init (pstr);
@@ -847,7 +850,7 @@ static void snd_pcm_free_stream(struct snd_pcm_str * pstr)#endif free_chmap (pstr);if (pstr->substream_count) -put_device (&pstr->dev); +put_device (pstr->dev); }#if IS_ENABLED (CONFIG_SND_PCM_OSS )
@@ -1017,7 +1020,7 @@ void snd_pcm_detach_substream(struct snd_pcm_substream *substream)static ssize_t pcm_class_show (struct device *dev,struct device_attribute *attr,char *buf) { - struct snd_pcm_str *pstr = container_of(dev, struct snd_pcm_str, dev); +struct snd_pcm_str *pstr =dev_get_drvdata (dev);struct snd_pcm *pcm = pstr->pcm;const char *str;static const char *strs[SNDRV_PCM_CLASS_LAST +1 ] = {
@@ -1078,7 +1081,7 @@ static int snd_pcm_dev_register(struct snd_device *device)/* register pcm */ err =snd_register_device (devtype, pcm->card, pcm->device, &snd_pcm_f_ops[cidx], pcm, - &pcm->streams[cidx].dev); + pcm->streams[cidx].dev);if (err <0 ) {list_del_init (&pcm->list); goto unlock;
@@ -1125,7 +1128,8 @@ static int snd_pcm_dev_disconnect(struct snd_device *device)pcm_call_notify (pcm, n_disconnect);for (cidx =0 ; cidx <2 ; cidx++) { -snd_unregister_device (&pcm->streams[cidx].dev); +if (pcm->streams[cidx].dev) +snd_unregister_device (pcm->streams[cidx].dev);free_chmap (&pcm->streams[cidx]); }mutex_unlock (&pcm->open_mutex);
sound/usb/media.c +1 -1
@@ -35,7 +35,7 @@ int snd_media_stream_init(struct snd_usb_substream *subs, struct snd_pcm *pcm, {struct media_device *mdev;struct media_ctl *mctl; -struct device *pcm_dev = &pcm->streams[stream].dev; +struct device *pcm_dev = pcm->streams[stream].dev;u32 intf_type;int ret =0 ;u16 mixer_pad;
Signature · 1 line
2.35.3
RFC PATCH 4/9 ALSA: hwdep: Don't embed device
In reply to Takashi IwaiLike control and PCM devices, it's better to avoid the embedded struct device for hwdep (although it's more or less well working), too. Change it to allocate via snd_device_alloc(), and free the memory at the common snd_hwdep_free().
--- include/sound/hwdep.h | 2 +- sound/core/hwdep.c | 38 +++++++++++++++++++++----------------- sound/pci/hda/hda_hwdep.c | 4 ++-- 3 files changed, 24 insertions(+), 20 deletions(-)
include/sound/hwdep.h +1 -1
@@ -53,7 +53,7 @@ struct snd_hwdep {wait_queue_head_t open_wait;void *private_data;void (*private_free) (struct snd_hwdep *hwdep); -struct device dev; +struct device *dev;struct mutex open_mutex;int used;/* reference counter */
sound/core/hwdep.c +21 -17
@@ -338,9 +338,14 @@ static const struct file_operations snd_hwdep_f_ops = .mmap = snd_hwdep_mmap, }; -static void release_hwdep_device (struct device *dev) +static void snd_hwdep_free (struct snd_hwdep *hwdep) { -kfree (container_of(dev ,struct snd_hwdep ,dev )); +if (!hwdep) +return ; +if (hwdep->private_free) + hwdep->private_free (hwdep); +put_device (hwdep->dev); +kfree (hwdep); }/**
@@ -382,16 +387,20 @@ int snd_hwdep_new(struct snd_card *card, char *id, int device,if (id)strscpy (hwdep->id, id,sizeof (hwdep->id)); -snd_device_initialize (&hwdep->dev, card); - hwdep->dev.release = release_hwdep_device; -dev_set_name (&hwdep->dev,"hwC%iD%i" , card->number, device); + err =snd_device_alloc (&hwdep->dev, card); +if (err <0 ) { +snd_hwdep_free (hwdep); +return err; + } + +dev_set_name (hwdep->dev,"hwC%iD%i" , card->number, device);#ifdef CONFIG_SND_OSSEMUL hwdep->oss_type =-1 ;#endif err =snd_device_new (card,SNDRV_DEV_HWDEP , hwdep, &ops);if (err <0 ) { -put_device (&hwdep->dev); +snd_hwdep_free (hwdep);return err; }
@@ -403,12 +412,7 @@ EXPORT_SYMBOL(snd_hwdep_new);static int snd_hwdep_dev_free (struct snd_device *device) { -struct snd_hwdep *hwdep = device->device_data; -if (!hwdep) -return 0 ; -if (hwdep->private_free) - hwdep->private_free (hwdep); -put_device (&hwdep->dev); +snd_hwdep_free (device->device_data);return 0 ; }
@@ -426,9 +430,9 @@ static int snd_hwdep_dev_register(struct snd_device *device)list_add_tail (&hwdep->list, &snd_hwdep_devices); err =snd_register_device (SNDRV_DEVICE_TYPE_HWDEP , hwdep->card, hwdep->device, - &snd_hwdep_f_ops, hwdep, &hwdep->dev); + &snd_hwdep_f_ops, hwdep, hwdep->dev);if (err <0 ) { -dev_err (&hwdep->dev,"unable to register\n" ); +dev_err (hwdep->dev,"unable to register\n" );list_del (&hwdep->list);mutex_unlock (®ister_mutex);return err;
@@ -439,12 +443,12 @@ static int snd_hwdep_dev_register(struct snd_device *device)if (hwdep->oss_type >=0 ) {if (hwdep->oss_type ==SNDRV_OSS_DEVICE_TYPE_DMFM && hwdep->device) -dev_warn (&hwdep->dev, +dev_warn (hwdep->dev,"only hwdep device 0 can be registered as OSS direct FM device!\n" );else if (snd_register_oss_device (hwdep->oss_type, card, hwdep->device, &snd_hwdep_f_ops, hwdep) <0 ) -dev_warn (&hwdep->dev, +dev_warn (hwdep->dev,"unable to register OSS compatibility device\n" );else hwdep->ossreg =1 ;
@@ -471,7 +475,7 @@ static int snd_hwdep_dev_disconnect(struct snd_device *device)if (hwdep->ossreg)snd_unregister_oss_device (hwdep->oss_type, hwdep->card, hwdep->device);#endif -snd_unregister_device (&hwdep->dev); +snd_unregister_device (hwdep->dev);list_del_init (&hwdep->list);mutex_unlock (&hwdep->open_mutex);mutex_unlock (®ister_mutex);
sound/pci/hda/hda_hwdep.c +2 -2
@@ -114,8 +114,8 @@ int snd_hda_create_hwdep(struct hda_codec *codec)#endif /* for sysfs */ - hwdep->dev.groups = snd_hda_dev_attr_groups; -dev_set_drvdata (&hwdep->dev, codec); + hwdep->dev->groups = snd_hda_dev_attr_groups; +dev_set_drvdata (hwdep->dev, codec);return 0 ; }
Signature · 1 line
2.35.3
RFC PATCH 5/9 ALSA: rawmidi: Don't embed device
In reply to Takashi IwaiThis patch detaches the struct device from the snd_rawmidi object by allocating via snd_device_alloc(), just like done for other devices.
--- include/sound/rawmidi.h | 2 +- sound/core/rawmidi.c | 29 +++++++++++++---------------- sound/core/ump.c | 8 ++++---- 3 files changed, 18 insertions(+), 21 deletions(-)
include/sound/rawmidi.h +1 -1
@@ -135,7 +135,7 @@ struct snd_rawmidi {struct mutex open_mutex;wait_queue_head_t open_wait; -struct device dev; +struct device *dev;struct snd_info_entry *proc_entry;
sound/core/rawmidi.c +13 -16
@@ -44,11 +44,11 @@ static LIST_HEAD(snd_rawmidi_devices);static DEFINE_MUTEX (register_mutex); #define rmidi_err(rmidi, fmt, args...) \ - dev_err(&(rmidi)->dev, fmt, ##args) + dev_err((rmidi)->dev, fmt, ##args) #define rmidi_warn(rmidi, fmt, args...) \ - dev_warn(&(rmidi)->dev, fmt, ##args) + dev_warn((rmidi)->dev, fmt, ##args) #define rmidi_dbg(rmidi, fmt, args...) \ - dev_dbg(&(rmidi)->dev, fmt, ##args) + dev_dbg((rmidi)->dev, fmt, ##args)struct snd_rawmidi_status32 {s32 stream;
@@ -1877,11 +1877,6 @@ static int snd_rawmidi_alloc_substreams(struct snd_rawmidi *rmidi,return 0 ; } -static void release_rawmidi_device (struct device *dev) -{ -kfree (container_of(dev ,struct snd_rawmidi ,dev )); -} -/* used for both rawmidi and ump */ int snd_rawmidi_init(struct snd_rawmidi *rmidi, struct snd_card *card, char *id, int device,
@@ -1906,12 +1901,13 @@ int snd_rawmidi_init(struct snd_rawmidi *rmidi,if (id !=NULL )strscpy (rmidi->id, id,sizeof (rmidi->id)); -snd_device_initialize (&rmidi->dev, card); - rmidi->dev.release = release_rawmidi_device; + err =snd_device_alloc (&rmidi->dev, card); +if (err <0 ) +return err;if (rawmidi_is_ump (rmidi)) -dev_set_name (&rmidi->dev,"umpC%iD%i" , card->number, device); +dev_set_name (rmidi->dev,"umpC%iD%i" , card->number, device);else -dev_set_name (&rmidi->dev,"midiC%iD%i" , card->number, device); +dev_set_name (rmidi->dev,"midiC%iD%i" , card->number, device); err = snd_rawmidi_alloc_substreams(rmidi, &rmidi->streams[SNDRV_RAWMIDI_STREAM_INPUT],
@@ -1996,7 +1992,8 @@ int snd_rawmidi_free(struct snd_rawmidi *rmidi)snd_rawmidi_free_substreams (&rmidi->streams[SNDRV_RAWMIDI_STREAM_OUTPUT ]);if (rmidi->private_free) rmidi->private_free (rmidi); -put_device (&rmidi->dev); +put_device (rmidi->dev); +kfree (rmidi);return 0 ; }EXPORT_SYMBOL_GPL (snd_rawmidi_free);
@@ -2038,7 +2035,7 @@ static int snd_rawmidi_dev_register(struct snd_device *device) err =snd_register_device (SNDRV_DEVICE_TYPE_RAWMIDI , rmidi->card, rmidi->device, - &snd_rawmidi_f_ops, rmidi, &rmidi->dev); + &snd_rawmidi_f_ops, rmidi, rmidi->dev);if (err <0 ) {rmidi_err (rmidi,"unable to register\n" ); goto error;
@@ -2103,7 +2100,7 @@ static int snd_rawmidi_dev_register(struct snd_device *device)return 0 ; error_unregister: -snd_unregister_device (&rmidi->dev); +snd_unregister_device (rmidi->dev); error:mutex_lock (®ister_mutex);list_del (&rmidi->list);
@@ -2142,7 +2139,7 @@ static int snd_rawmidi_dev_disconnect(struct snd_device *device) rmidi->ossreg = 0; }#endif /* CONFIG_SND_OSSEMUL */ -snd_unregister_device (&rmidi->dev); +snd_unregister_device (rmidi->dev);mutex_unlock (&rmidi->open_mutex);mutex_unlock (®ister_mutex);return 0 ;
sound/core/ump.c +4 -4
@@ -13,10 +13,10 @@#include <sound/ump.h> #include <sound/ump_convert.h> -#define ump_err(ump, fmt, args...) dev_err(&(ump)->core.dev, fmt, ##args) -#define ump_warn(ump, fmt, args...) dev_warn(&(ump)->core.dev, fmt, ##args) -#define ump_info(ump, fmt, args...) dev_info(&(ump)->core.dev, fmt, ##args) -#define ump_dbg(ump, fmt, args...) dev_dbg(&(ump)->core.dev, fmt, ##args) +#define ump_err(ump, fmt, args...) dev_err((ump)->core.dev, fmt, ##args) +#define ump_warn(ump, fmt, args...) dev_warn((ump)->core.dev, fmt, ##args) +#define ump_info(ump, fmt, args...) dev_info((ump)->core.dev, fmt, ##args) +#define ump_dbg(ump, fmt, args...) dev_dbg((ump)->core.dev, fmt, ##args)static int snd_ump_dev_register (struct snd_rawmidi *rmidi);static int snd_ump_dev_unregister (struct snd_rawmidi *rmidi);
Signature · 1 line
2.35.3
RFC PATCH 6/9 ALSA: compress: Don't embed device
In reply to Takashi IwaiEmbedding the struct device to snd_compr object may result in UAF when the delayed kobj release is used. Like other devices, let's detach the struct device from the snd_compr by allocating dynamically via snd_device_alloc().
--- include/sound/compress_driver.h | 2 +- sound/core/compress_offload.c | 16 ++++++++++------ 2 files changed, 11 insertions(+), 7 deletions(-)
include/sound/compress_driver.h +1 -1
@@ -148,7 +148,7 @@ struct snd_compr_ops {*/ struct snd_compr {const char *name; -struct device dev; +struct device *dev;struct snd_compr_ops *ops;void *private_data;struct snd_card *card;
sound/core/compress_offload.c +10 -6
@@ -546,7 +546,7 @@ static int snd_compr_allocate_buffer(struct snd_compr_stream *stream,if (stream->runtime->dma_buffer_p) {if (buffer_size > stream->runtime->dma_buffer_p->bytes) -dev_err (&stream->device->dev, +dev_err (stream->device->dev,"Not enough DMA buffer" );else buffer = stream->runtime->dma_buffer_p->area;
@@ -1070,7 +1070,7 @@ static int snd_compress_dev_register(struct snd_device *device)/* register compressed device */ ret =snd_register_device (SNDRV_DEVICE_TYPE_COMPRESS , compr->card, compr->device, - &snd_compr_file_ops, compr, &compr->dev); + &snd_compr_file_ops, compr, compr->dev);if (ret <0 ) {pr_err ("snd_register_device failed %d\n" , ret);return ret;
@@ -1084,7 +1084,7 @@ static int snd_compress_dev_disconnect(struct snd_device *device)struct snd_compr *compr; compr = device->device_data; -snd_unregister_device (&compr->dev); +snd_unregister_device (compr->dev);return 0 ; }
@@ -1158,7 +1158,7 @@ static int snd_compress_dev_free(struct snd_device *device) compr = device->device_data;snd_compress_proc_done (compr); -put_device (&compr->dev); +put_device (compr->dev);return 0 ; }
@@ -1189,12 +1189,16 @@ int snd_compress_new(struct snd_card *card, int device,snd_compress_set_id (compr, id); -snd_device_initialize (&compr->dev, card); -dev_set_name (&compr->dev,"comprC%iD%i" , card->number, device); + ret =snd_device_alloc (&compr->dev, card); +if (ret) +return ret; +dev_set_name (compr->dev,"comprC%iD%i" , card->number, device); ret =snd_device_new (card,SNDRV_DEV_COMPRESS , compr, &ops);if (ret ==0 )snd_compress_proc_init (compr); +else +put_device (compr->dev);return ret; }
Signature · 1 line
2.35.3
RFC PATCH 7/9 ALSA: timer: Create device with snd_device_alloc()
In reply to Takashi IwaiAlign with the other components, and use snd_device_alloc() for the new sound device for timer, too. No functional changes.
--- sound/core/timer.c | 16 +++++++++------- 1 file changed, 9 insertions(+), 7 deletions(-)
sound/core/timer.c +9 -7
@@ -2301,7 +2301,7 @@ static void snd_timer_free_all(void)snd_timer_free (timer); } -static struct device timer_dev; +static struct device *timer_dev;/* * ENTRY functions
@@ -2311,8 +2311,10 @@ static int __init alsa_timer_init(void) {int err; -snd_device_initialize (&timer_dev,NULL ); -dev_set_name (&timer_dev,"timer" ); + err =snd_device_alloc (&timer_dev,NULL ); +if (err <0 ) +return err; +dev_set_name (timer_dev,"timer" );#ifdef SNDRV_OSS_INFO_DEV_TIMERS snd_oss_info_register(SNDRV_OSS_INFO_DEV_TIMERS, SNDRV_CARDS - 1,
@@ -2326,7 +2328,7 @@ static int __init alsa_timer_init(void) } err =snd_register_device (SNDRV_DEVICE_TYPE_TIMER ,NULL ,0 , - &snd_timer_f_ops,NULL , &timer_dev); + &snd_timer_f_ops,NULL , timer_dev);if (err <0 ) {pr_err ("ALSA: unable to register timer device (%i)\n" , err);snd_timer_free_all ();
@@ -2337,15 +2339,15 @@ static int __init alsa_timer_init(void)return 0 ; put_timer: -put_device (&timer_dev); +put_device (timer_dev);return err; } static void __exit alsa_timer_exit(void) { -snd_unregister_device (&timer_dev); +snd_unregister_device (timer_dev);snd_timer_free_all (); -put_device (&timer_dev); +put_device (timer_dev);snd_timer_proc_done ();#ifdef SNDRV_OSS_INFO_DEV_TIMERS snd_oss_info_unregister (SNDRV_OSS_INFO_DEV_TIMERS ,SNDRV_CARDS -1 );
Signature · 1 line
2.35.3
RFC PATCH 8/9 ALSA: seq: Create device with snd_device_alloc()
In reply to Takashi IwaiAlign with the other components, and use snd_device_alloc() for the new sound device for sequencer, too. No functional changes.
--- sound/core/seq/seq_clientmgr.c | 16 +++++++++------- 1 file changed, 9 insertions(+), 7 deletions(-)
sound/core/seq/seq_clientmgr.c +9 -7
@@ -2721,7 +2721,7 @@ static const struct file_operations snd_seq_f_ops = .compat_ioctl = snd_seq_ioctl_compat, }; -static struct device seq_dev; +static struct device *seq_dev;/* * register sequencer device
@@ -2730,15 +2730,17 @@ int __init snd_sequencer_device_init(void) {int err; -snd_device_initialize (&seq_dev,NULL ); -dev_set_name (&seq_dev,"seq" ); + err =snd_device_alloc (&seq_dev,NULL ); +if (err <0 ) +return err; +dev_set_name (seq_dev,"seq" );mutex_lock (®ister_mutex); err =snd_register_device (SNDRV_DEVICE_TYPE_SEQUENCER ,NULL ,0 , - &snd_seq_f_ops,NULL , &seq_dev); + &snd_seq_f_ops,NULL , seq_dev);mutex_unlock (®ister_mutex);if (err <0 ) { -put_device (&seq_dev); +put_device (seq_dev);return err; }
@@ -2752,6 +2754,6 @@ int __init snd_sequencer_device_init(void)*/ void snd_sequencer_device_done (void ) { -snd_unregister_device (&seq_dev); -put_device (&seq_dev); +snd_unregister_device (seq_dev); +put_device (seq_dev); }
Signature · 1 line
2.35.3
RFC PATCH 9/9 ALSA: core: Drop snd_device_initialize()
In reply to Takashi IwaiNow all users of snd_device_intialize() are gone, let's drop it.
--- include/sound/core.h | 1 - sound/core/init.c | 23 ----------------------- 2 files changed, 24 deletions(-)
include/sound/core.h +0 -1
@@ -240,7 +240,6 @@ extern struct dentry *sound_debugfs_root;void snd_request_card (int card);int snd_device_alloc (struct device **dev_p,struct snd_card *card); -void snd_device_initialize (struct device *dev,struct snd_card *card); int snd_register_device(int type, struct snd_card *card, int dev, const struct file_operations *f_ops,
sound/core/init.c +0 -23
@@ -111,29 +111,6 @@ static int get_slot_from_bitmask(int mask, int (*check)(struct module *, int),return mask;/* unchanged */ } -/* the default release callback set in snd_device_initialize() below; -* this is just NOP for now, as almost all jobs are already done in -* dev_free callback of snd_device chain instead. -*/ -static void default_release (struct device *dev) -{ -} - -/** -* snd_device_initialize - Initialize struct device for sound devices -* @dev: device to initialize -* @card: card to assign, optional -*/ -void snd_device_initialize (struct device *dev,struct snd_card *card) -{ -device_initialize (dev); -if (card) - dev->parent = &card->card_dev; - dev->class = &sound_class; - dev->release = default_release; -} -EXPORT_SYMBOL_GPL (snd_device_initialize); -/* the default release callback set in snd_device_alloc() */ static void default_release_alloc (struct device *dev) {
Signature · 1 line
2.35.3
Quoted 34 lines · Takashi Iwai
Hi,
this is another set of patches to attempt papering over the UAF problems that are seen when the delayed kobject release is enabled, as initially reported by Curtis:
https://lore.kernel.org/r/[email protected]There was a previous patch set with a different approach (using the device refcount dependencies), but this is a sort of step-back to the old way.
https://lore.kernel.org/r/[email protected]After discussions and evaluations, we agreed that decoupling the struct device from each sound component object is the safest (and easiest) way as of now. For applying the changes more consistently, I introduced a new helper for the struct device allocation and initialization, and applied all components.
A couple of more changes for
card_devrefcount managed aren't included in this patch set, though. They might be good to have, but this patch set should suffice for the currently seen UAF problems.For a long-term solution, we may restructure the device management, then the struct devices may be embedded again in each object. But, it'll need lots of other changes and cleanups, a big TODO.
The latest patches are found in topic/dev-split branch of sound.git tree.
Takashi
===
Quoted 10 lines
Takashi Iwai (9):
ALSA: core: Introducesnd_device_alloc()
ALSA: control: Don't embedctl_dev
ALSA: pcm: Don't embed device
ALSA: hwdep: Don't embed device
ALSA: rawmidi: Don't embed device
ALSA: compress: Don't embed device
ALSA: timer: Create device withsnd_device_alloc()
ALSA: seq: Create device withsnd_device_alloc()
ALSA: core: Dropsnd_device_initialize()
For all commits:
Signature · 2 lines
Jaroslav Kysela <[email protected]> Linux Sound Maintainer; ALSA Project; Red Hat, Inc.
Curtis Malainey
Tested-byQuoted 20 lines · Jaroslav Kysela
Quoted 18 lines · Takashi Iwai
Hi,
this is another set of patches to attempt papering over the UAF problems that are seen when the delayed kobject release is enabled, as initially reported by Curtis:
https://lore.kernel.org/r/[email protected]There was a previous patch set with a different approach (using the device refcount dependencies), but this is a sort of step-back to the old way.
https://lore.kernel.org/r/[email protected]After discussions and evaluations, we agreed that decoupling the struct device from each sound component object is the safest (and easiest) way as of now. For applying the changes more consistently, I introduced a new helper for the struct device allocation and initialization, and applied all components.
Quoted 7 lines
Quoted 7 lines
A couple of more changes for
card_devrefcount managed aren't included in this patch set, though. They might be good to have, but this patch set should suffice for the currently seen UAF problems.For a long-term solution, we may restructure the device management, then the struct devices may be embedded again in each object. But, it'll need lots of other changes and cleanups, a big TODO.
I agree I think we should apply this as proper fixes will be a big lift. Thanks for refining them.
Quoted 20 lines
Quoted 9 lines
The latest patches are found in topic/dev-split branch of sound.git tree.
Takashi
===
Quoted 10 lines
Takashi Iwai (9):
ALSA: core: Introducesnd_device_alloc()
ALSA: control: Don't embedctl_dev
ALSA: pcm: Don't embed device
ALSA: hwdep: Don't embed device
ALSA: rawmidi: Don't embed device
ALSA: compress: Don't embed device
ALSA: timer: Create device withsnd_device_alloc()
ALSA: seq: Create device withsnd_device_alloc()
ALSA: core: Dropsnd_device_initialize()
Quoted 8 lines
For all commits:
Reviewed-by Jaroslav Kysela <[email protected]>--
Jaroslav Kysela <[email protected]>
Linux Sound Maintainer; ALSA Project; Red Hat, Inc.
For all
Quoted 35 lines · Takashi Iwai
Hi,
this is another set of patches to attempt papering over the UAF problems that are seen when the delayed kobject release is enabled, as initially reported by Curtis:
https://lore.kernel.org/r/[email protected]There was a previous patch set with a different approach (using the device refcount dependencies), but this is a sort of step-back to the old way.
https://lore.kernel.org/r/[email protected]After discussions and evaluations, we agreed that decoupling the struct device from each sound component object is the safest (and easiest) way as of now. For applying the changes more consistently, I introduced a new helper for the struct device allocation and initialization, and applied all components.
A couple of more changes for
card_devrefcount managed aren't included in this patch set, though. They might be good to have, but this patch set should suffice for the currently seen UAF problems.For a long-term solution, we may restructure the device management, then the struct devices may be embedded again in each object. But, it'll need lots of other changes and cleanups, a big TODO.
The latest patches are found in topic/dev-split branch of sound.git tree.
Takashi
===
Quoted 10 lines
Takashi Iwai (9):
ALSA: core: Introducesnd_device_alloc()
ALSA: control: Don't embedctl_dev
ALSA: pcm: Don't embed device
ALSA: hwdep: Don't embed device
ALSA: rawmidi: Don't embed device
ALSA: compress: Don't embed device
ALSA: timer: Create device withsnd_device_alloc()
ALSA: seq: Create device withsnd_device_alloc()
ALSA: core: Dropsnd_device_initialize()
Although the patch set was sent as RFC, I merged them now for 6.6 with Acks, as there is no further plan to change.
thanks,
Takashi