Currently, kernel/module annotates module dependencies when request_symbol is used, but it doesn't cover more complex inter-driver dependencies that are subsystem and/or driver-specific.
In the case of hdmi sound, depending on the CPU/GPU, sometimes the snd_hda_driver can talk directly with the hardware, but sometimes, it uses the i915 driver. When the snd_hda_driver uses i915, it should first be unbind/rmmod, as otherwise trying to unbind/rmmod the i915 driver cause driver issues, as as reported by CI tools with different GPU models:
https://intel-gfx-ci.01.org/tree/drm-tip/IGT_6415/fi-tgl-1115g4/igt@core_hot... https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_11495/bat-adlm-1/igt@i915_mo...
In the past, just a few CPUs were doing such bindings, but this issue now applies to all "modern" Intel CPUs that have onboard graphics, as well as to the newer discrete GPUs.
With the discrete GPU case, the HDA controller is physically separate and requires i915 to power on the hardware for all hardware access. In this case, the issue is hit basicly 100% of the time.
With on-board graphics, i915 driver is needed only when the display codec is accessed. If i915 is unbind during runtime suspend, while snd-hda-intel is still bound, nothing bad happens, but unbinding i915 on other situations may also cause issues.
So, add support at kernel/modules to allow snd-hda drivers to properly annotate when a dependency on a DRM driver dependencies exists, and add a call to such new function at the snd-hda driver when it successfully binds into the DRM driver.
This would allow userspace tools to check and properly remove the audio driver before trying to remove or unbind the GPU driver.
It should be noticed that this series conveys the hidden module dependencies. Other changes are needed in order to allow removing or unbinding the i915 driver while keeping the snd-hda-intel driver loaded/bound. With that regards, there are some discussions on how to improve this at alsa-devel a while back:
https://mailman.alsa-project.org/pipermail/alsa-devel/2021-September/190099....
So, future improvements on both in i915 and the audio drivers could be made. E.g. with discrete GPUs, it's the only codec of the card, so it seems feasible to detach the ALSA card if i915 is bound (using infra made for VGA switcheroo), but, until these improvements are done and land in upstream, audio drivers needs to be unbound if i915 driver goes unbind.
Yet, even if such fixes got merged, this series is still needed, as it makes such dependencies more explicit and easier to debug.
PS.: This series was generated against next-20220428.
---
v2: the dependencies are now handled directly at try_module_get().
Mauro Carvalho Chehab (2): module: update dependencies at try_module_get() ALSA: hda - identify when audio is provided by a video driver
include/linux/module.h | 4 +++- kernel/module/main.c | 35 +++++++++++++++++++++++++++++++++-- sound/hda/hdac_component.c | 2 +- 3 files changed, 37 insertions(+), 4 deletions(-)
Sometimes, device drivers are bound into each other via try_module_get(), making such references invisible when looking at /proc/modules or lsmod.
Add a function to allow setting up module references for such cases, and call it when try_module_get() is used.
Reviewed-by: Dan Williams dan.j.williams@intel.com Signed-off-by: Mauro Carvalho Chehab mchehab@kernel.org ---
See [PATCH v2 0/2] at: https://lore.kernel.org/all/cover.1651314499.git.mchehab@kernel.org/
include/linux/module.h | 4 +++- kernel/module/main.c | 35 +++++++++++++++++++++++++++++++++-- 2 files changed, 36 insertions(+), 3 deletions(-)
diff --git a/include/linux/module.h b/include/linux/module.h index 46d4d5f2516e..836851baaad4 100644 --- a/include/linux/module.h +++ b/include/linux/module.h @@ -620,7 +620,9 @@ extern void __module_get(struct module *module);
/* This is the Right Way to get a module: if it fails, it's being removed, * so pretend it's not there. */ -extern bool try_module_get(struct module *module); +extern bool __try_module_get(struct module *module, struct module *this); + +#define try_module_get(mod) __try_module_get(mod, THIS_MODULE)
extern void module_put(struct module *module);
diff --git a/kernel/module/main.c b/kernel/module/main.c index 05a42d8fcd7a..9f4416381e65 100644 --- a/kernel/module/main.c +++ b/kernel/module/main.c @@ -631,6 +631,35 @@ static int ref_module(struct module *a, struct module *b) return 0; }
+static int ref_module_dependency(struct module *mod, + struct module *this) +{ + int ret; + + if (!this || !this->name) { + return -EINVAL; + } + + if (mod == this) + return 0; + + mutex_lock(&module_mutex); + + ret = ref_module(this, mod); + +#ifdef CONFIG_MODULE_UNLOAD + if (ret) + goto ret; + + ret = sysfs_create_link(mod->holders_dir, + &this->mkobj.kobj, this->name); +#endif + +ret: + mutex_unlock(&module_mutex); + return ret; +} + /* Clear the unload stuff of the module. */ static void module_unload_free(struct module *mod) { @@ -841,7 +870,7 @@ void __module_get(struct module *module) } EXPORT_SYMBOL(__module_get);
-bool try_module_get(struct module *module) +bool __try_module_get(struct module *module, struct module *this) { bool ret = true;
@@ -856,9 +885,11 @@ bool try_module_get(struct module *module)
preempt_enable(); } + if (ret) + ref_module_dependency(module, this); return ret; } -EXPORT_SYMBOL(try_module_get); +EXPORT_SYMBOL(__try_module_get);
void module_put(struct module *module) {
On Sat, Apr 30, 2022 at 11:30:58AM +0100, Mauro Carvalho Chehab wrote:
Sometimes, device drivers are bound into each other via try_module_get(), making such references invisible when looking at /proc/modules or lsmod.
Add a function to allow setting up module references for such cases, and call it when try_module_get() is used.
Reviewed-by: Dan Williams dan.j.williams@intel.com Signed-off-by: Mauro Carvalho Chehab mchehab@kernel.org
See [PATCH v2 0/2] at: https://lore.kernel.org/all/cover.1651314499.git.mchehab@kernel.org/
include/linux/module.h | 4 +++- kernel/module/main.c | 35 +++++++++++++++++++++++++++++++++-- 2 files changed, 36 insertions(+), 3 deletions(-)
diff --git a/include/linux/module.h b/include/linux/module.h index 46d4d5f2516e..836851baaad4 100644 --- a/include/linux/module.h +++ b/include/linux/module.h @@ -620,7 +620,9 @@ extern void __module_get(struct module *module);
/* This is the Right Way to get a module: if it fails, it's being removed,
- so pretend it's not there. */
-extern bool try_module_get(struct module *module); +extern bool __try_module_get(struct module *module, struct module *this);
+#define try_module_get(mod) __try_module_get(mod, THIS_MODULE)
extern void module_put(struct module *module);
diff --git a/kernel/module/main.c b/kernel/module/main.c index 05a42d8fcd7a..9f4416381e65 100644 --- a/kernel/module/main.c +++ b/kernel/module/main.c @@ -631,6 +631,35 @@ static int ref_module(struct module *a, struct module *b) return 0; }
+static int ref_module_dependency(struct module *mod,
struct module *this)
This can be on one line, right?
+{
- int ret;
- if (!this || !this->name) {
return -EINVAL;- }
Did you run checkpatch on this? Please do :)
- if (mod == this)
return 0;
How can this happen?
When people mistakenly call try_module_get(THIS_MODULE)? We should throw up a big warning when that happens anyway as that's always wrong.
But that's a different issue from this change, sorry for the noise.
- mutex_lock(&module_mutex);
- ret = ref_module(this, mod);
+#ifdef CONFIG_MODULE_UNLOAD
- if (ret)
goto ret;- ret = sysfs_create_link(mod->holders_dir,
&this->mkobj.kobj, this->name);
Meta comment, why do we only create links if we can unload things?
thanks,
greg k-h
Em Sat, 30 Apr 2022 14:04:59 +0200 Greg KH gregkh@linuxfoundation.org escreveu:
On Sat, Apr 30, 2022 at 11:30:58AM +0100, Mauro Carvalho Chehab wrote:
Did you run checkpatch on this? Please do :)
- if (mod == this)
return 0;How can this happen? When people mistakenly call try_module_get(THIS_MODULE)?
Yes. There are lots of place where this is happening:
$ git grep try_module_get(THIS_MODULE|wc -l 82
We should throw up a big warning when that happens anyway as that's always wrong.
But that's a different issue from this change, sorry for the noise.
It sounds very weird to use try_module_get(THIS_MODULE).
We could add a WARN_ON() there - or something similar - but I would do it on a separate patch.
- mutex_lock(&module_mutex);
- ret = ref_module(this, mod);
+#ifdef CONFIG_MODULE_UNLOAD
- if (ret)
goto ret;- ret = sysfs_create_link(mod->holders_dir,
&this->mkobj.kobj, this->name);Meta comment, why do we only create links if we can unload things?
Good question. I don't know for certain. This is the already existing pattern at add_usage_links() - see kernel/module/sysfs.c.
Also, lsmod uses sysfs links when showing dependencies.
Regards, Mauro
From: Mauro Carvalho Chehab
Sent: 30 April 2022 14:38
Em Sat, 30 Apr 2022 14:04:59 +0200 Greg KH gregkh@linuxfoundation.org escreveu:
On Sat, Apr 30, 2022 at 11:30:58AM +0100, Mauro Carvalho Chehab wrote:
Did you run checkpatch on this? Please do :)
- if (mod == this)
return 0;How can this happen? When people mistakenly call try_module_get(THIS_MODULE)?
Yes. There are lots of place where this is happening:
$ git grep try_module_get(THIS_MODULE|wc -l 82
We should throw up a big warning when that happens anyway as that's always wrong.
But that's a different issue from this change, sorry for the noise.
It sounds very weird to use try_module_get(THIS_MODULE).
We could add a WARN_ON() there - or something similar - but I would do it on a separate patch.
You could add a compile-time check. But a run-time one seems unnecessary. Clearly try_module_get(THIS_MODULE) usually succeeds.
I think I can invent a case where it can fail: The module count must be zero, and a module unload in progress. The thread doing the unload is blocked somewhere. Another thread makes a callback into the module for some request that (for instance) would need to create a kernel thread. It tries to get a reference for the thread. So try_module_get(THIS_MODULE) is the right call - and will fail here.
David
- Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)
On some devices, the hda driver needs to hook into a video driver, in order to be able to properly access the audio hardware and/or the power management function.
That's the case of several snd_hda_intel devices that depends on i915 driver.
Ensure that a proper reference between the snd-hda driver needing such binding is shown at /proc/modules, in order to allow userspace to know about such binding.
Signed-off-by: Mauro Carvalho Chehab mchehab@kernel.org ---
See [PATCH v2 0/2] at: https://lore.kernel.org/all/cover.1651314499.git.mchehab@kernel.org/
sound/hda/hdac_component.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/sound/hda/hdac_component.c b/sound/hda/hdac_component.c index bb37e7e0bd79..30e130457272 100644 --- a/sound/hda/hdac_component.c +++ b/sound/hda/hdac_component.c @@ -199,7 +199,7 @@ static int hdac_component_master_bind(struct device *dev) }
/* pin the module to avoid dynamic unbinding, but only if given */ - if (!try_module_get(acomp->ops->owner)) { + if (!__try_module_get(acomp->ops->owner, dev->driver->owner)) { ret = -ENODEV; goto out_unbind; }
dri-devel@lists.freedesktop.org
-
David Laight -
Greg KH -
Mauro Carvalho Chehab