Skip to contentSkip to conversation

snd-usb-usx2y 0.8.0 against alsa-kernel cvshead: OHCI doesn't crash anymore

PATCH In review Karsten Wiese 4 participants 16 messages Sep 30, 2004 – Oct 26, 2004

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 snd-usb-usx2y 0.8.0 against alsa-kernel cvshead: OHCI doesn't crash anymore

Summary: Prevents crashes when communicating over ohci usb-hcds.

Work on this started, when rumors spread that ohci equipped machines would crash.
This was due to me missing two facts:
1) Ohci has a bigger usb frame number wrap around.
2) It only supports URB_ISO_ASAP when submitting iso urbs. Fixed now.
Usb-traffic for playback starts now in the prepare callback, like it previously only did for capture.
As a side effect code became simpler and smaller.

#1 lore raw

At Thu, 30 Sep 2004 20:46:29 +0200,

Quoted 13 lines · Karsten Wiese

Summary: Prevents crashes when communicating over ohci usb-hcds.

Work on this started, when rumors spread that ohci equipped machines would crash.
This was due to me missing two facts:
1) Ohci has a bigger usb frame number wrap around.
2) It only supports URB_ISO_ASAP when submitting iso urbs. Fixed now.
Usb-traffic for playback starts now in the prepare callback, like it previously only did for capture.
As a side effect code became simpler and smaller.

Quoted 1 line

Thanks for the patch.

What is the purpose of the replacement with USX2Y_STATIC?

One thing I'm not sure is that the possible race of usX2Y_urbs_start() and usX2Y_subs_startup_finish(). Better to use complete handler?

Takashi

-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal Use IT products in your business? Tell us what you think of them. Give us Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl

#2 lore raw

Am Freitag 01 Oktober 2004 15:50 schrieb Takashi Iwai:

Quoted 3 lines

What is the purpose of the replacement with USX2Y_STATIC?

Its for easier appliance of the patch, I sent a little later yesterday 22:01:50:
"[Alsa-devel] [PATCH]hwdep addon for raw usb mmap from userspace against snd-usb-usx2y 0.8.0 in alsa-kernel tree"
The addon needs the functions with the USX2Y_STATIC.
So the addon patch only needs to change

	#define USX2Y_STATIC static
to
	#define USX2Y_STATIC

.

Quoted 3 lines

One thing I'm not sure is that the possible race of usX2Y_urbs_start() and usX2Y_subs_startup_finish(). Better to use complete handler?

Thanks for the hint. Will look at it again.

Karsten

-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal Use IT products in your business? Tell us what you think of them. Give us Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl

#3 lore raw

At Fri, 1 Oct 2004 18:28:32 +0200,

Quoted 11 lines · Karsten Wiese

Am Freitag 01 Oktober 2004 15:50 schrieb Takashi Iwai:

Quoted 3 lines

What is the purpose of the replacement with USX2Y_STATIC?

Its for easier appliance of the patch, I sent a little later yesterday 22:01:50:
"[Alsa-devel] [PATCH]hwdep addon for raw usb mmap from userspace against snd-usb-usx2y 0.8.0 in alsa-kernel tree"
The addon needs the functions with the USX2Y_STATIC.
So the addon patch only needs to change

Quoted 4 lines
	#define USX2Y_STATIC static
to
	#define USX2Y_STATIC

.

Well, this won't make the patch applicance easier at all.

If the second patch weren't accepted, it's because the patch does really bad indeed (e.g. written too dirty or conceptually wrong). Then USX2Y_STATIC in the first patch would be useless, anyway.

If the second patch were accepted, you can include static -> extern conversion in this patch without problems. Thus, having USX2Y_STATIC in the first patch isn't necessary, too.

In sort: You shouldn't sacrifice the right patch just because of others :)

Takashi

-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal Use IT products in your business? Tell us what you think of them. Give us Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl

#4 lore raw

PATCH 1/1 snd-usb-usx2y 0.8.1 against alsa-kernel cvshead: OHCI doesn't crash anymore

In reply to Takashi Iwai

Am Freitag 01 Oktober 2004 15:50 schrieben Sie:

Quoted 1 line

What is the purpose of the replacement with USX2Y_STATIC?

gone now.

Quoted 3 lines

One thing I'm not sure is that the possible race of usX2Y_urbs_start() and usX2Y_subs_startup_finish(). Better to use complete handler?

did I care for any possible racing now?

Best,
Karsten

patch header:

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

Summary: Prevents crashes when communicating over ohci usb-hcds.

Work on this started, when rumors spread that ohci equipped machines would crash.
This was due to me missing two facts:
1) Ohci has a bigger usb frame number wrap around.
2) It only supports URB_ISO_ASAP when submitting iso urbs. Fixed now.
Usb-traffic for playback starts now in the prepare callback, like it previously only did for capture.
As a side effect code became simpler and smaller.

<<<<

#5 lore raw

Re: [PATCH]snd-usb-usx2y 0.8.1 against alsa-kernel cvshead: OHCI doesn't crash anymore

At Sun, 3 Oct 2004 00:38:04 +0200,

Quoted 5 lines · Karsten Wiese
Quoted 3 lines

One thing I'm not sure is that the possible race of usX2Y_urbs_start() and usX2Y_subs_startup_finish(). Better to use complete handler?

did I care for any possible racing now?

You need to call spin_lock(&usX2Y->prepare_lock) before checking subs->prepared in usX2Y_urbs_start(). In this kind of implementation, all the access to subs->prepared should be protected in the spinlock to avoid races.

Also, the spin_lock(&usX2Y->prepare_lock) in usX2Y_urbs_start() may cause deadlock. Use spin_lock_irq(&usX2y->prepare_lock) instead.

Thanks,

Takashi

-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal Use IT products in your business? Tell us what you think of them. Give us Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl

#6 lore raw

Re: [PATCH]snd-usb-usx2y 0.8.1 against alsa-kernel cvshead: OHCI doesn't crash anymore

Am Montag 04 Oktober 2004 18:22 schrieb Takashi Iwai:

Quoted 8 lines

At Sun, 3 Oct 2004 00:38:04 +0200,

Quoted 4 lines · Karsten Wiese
Quoted 2 lines

One thing I'm not sure is that the possible race of usX2Y_urbs_start() and usX2Y_subs_startup_finish(). Better to use complete handler?

did I care for any possible racing now?

Quoted 8 lines

You need to call spin_lock(&usX2Y->prepare_lock) before checking subs->prepared in usX2Y_urbs_start(). In this kind of implementation, all the access to subs->prepared should be protected in the spinlock to avoid races.

Also, the spin_lock(&usX2Y->prepare_lock) in usX2Y_urbs_start() may cause deadlock. Use spin_lock_irq(&usX2y->prepare_lock) instead.

Hi Takashi

No spinlock no more. atomic_t instead. Please check.

Rui, the error in your dmesg log: "Debug: sleeping function called from invalid context .... " should be gone with this.

thanks,
Karsten

patch header:

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

Summary: Fixes crashes that occured when communicating over ohci usb-hcds.

Work on this started, when rumors spread that ohci equipped machines would crash.
This was due to me missing two facts:
1) Ohci has a bigger usb frame number wrap around.
2) It only supports URB_ISO_ASAP when submitting iso urbs. Fixed now.

<<<<

#7 lore raw

Re: [PATCH]snd-usb-usx2y 0.8.1 against alsa-kernel cvshead: OHCI doesn't crash anymore

At Mon, 18 Oct 2004 14:06:43 +0200,

Quoted 16 lines · Karsten Wiese

[1 <text/plain; iso-8859-1 (7bit)>]
Am Montag 04 Oktober 2004 18:22 schrieb Takashi Iwai:

Quoted 13 lines

At Sun, 3 Oct 2004 00:38:04 +0200,

Quoted 4 lines · Karsten Wiese
Quoted 2 lines

One thing I'm not sure is that the possible race of usX2Y_urbs_start() and usX2Y_subs_startup_finish(). Better to use complete handler?

did I care for any possible racing now?

You need to call spin_lock(&usX2Y->prepare_lock) before checking subs->prepared in usX2Y_urbs_start(). In this kind of implementation, all the access to subs->prepared should be protected in the spinlock to avoid races.

Quoted 6 lines
Quoted 3 lines

Also, the spin_lock(&usX2Y->prepare_lock) in usX2Y_urbs_start() may cause deadlock. Use spin_lock_irq(&usX2y->prepare_lock) instead.

Hi Takashi

No spinlock no more. atomic_t instead. Please check.

I'm afraid it's still racy.
In usX2Y_urbs_start(), the complete callback may be called between usb_submit_urb() and the loop of schedule_timeout().
So, the loop should check at first the condition before sleeping like below:

	while (atomic_read(&subs->state) != state_PREPARED) {
		timeout = schedule_timeout(timeout);
		if (signal_pending(current))
			...
		if (! timeout)
			...
	}

Regarding the new config stuff:
Let's kill CONFIG_SND_USB_USX2Y_NRPACKS_VARIABLE option. No one will want to disable such a module option.

Also, please avoid the Kconfig int value as much as possible if you want to support the driver for 2.2/2.4 kernels, too. I believe it's ok to put a hardcoded value 4 like usbaudio.c if you describe the module option properly in the document.

Takashi

-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal Use IT products in your business? Tell us what you think of them. Give us Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl

#8 lore raw

Re: [PATCH]snd-usb-usx2y 0.8.1 against alsa-kernel cvshead: OHCI doesn't crash anymore

Am Dienstag 19 Oktober 2004 15:33 schrieb Takashi Iwai:

Quoted 7 lines

I'm afraid it's still racy.
In usX2Y_urbs_start(), the complete callback may be called between usb_submit_urb() and the loop of schedule_timeout().
So, the loop should check at first the condition before sleeping like below:

Quoted 8 lines
	while (atomic_read(&subs->state) != state_PREPARED) {
		timeout = schedule_timeout(timeout);
		if (signal_pending(current))
			...
		if (! timeout)
			...
	}

<click, beep, ring> ok, thanks!
But there is still a small race window between

	while (atomic_read(&subs->state) != state_PREPARED) {
and   /* right here  the complete callback may be called */
		timeout = schedule_timeout(timeout);

no?
Is there a trick to avoid even this small race window?

Quoted 5 lines

Regarding the new config stuff:
Let's kill CONFIG_SND_USB_USX2Y_NRPACKS_VARIABLE option. No one will want to disable such a module option.

Well, me wants to disable it as it works best with nrpacks=1 here and I'd like the machine code to not contain the unneeded loops then...
(gcc optimizes those loops away if it reads "#define nrpacks 1") Can I keep some (commented out?) define switch inside the source (invisible to KConfig) to suit my needs?

Quoted 5 lines

Also, please avoid the Kconfig int value as much as possible if you want to support the driver for 2.2/2.4 kernels, too. I believe it's ok to put a hardcoded value 4 like usbaudio.c if you describe the module option properly in the document.

Erm, which document?
And I'm not too keen to do the driver (sup/back)port for 2.2/2.4 kernels myself. I'll avoid the Kconfig int value, making it simpler for volunteers.

Karsten

-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal Use IT products in your business? Tell us what you think of them. Give us Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl

#9 lore raw

Re: [PATCH]snd-usb-usx2y 0.8.1 against alsa-kernel cvshead: OHCI doesn't crash anymore

At Wed, 20 Oct 2004 02:28:15 +0200,

Quoted 17 lines · Karsten Wiese

Am Dienstag 19 Oktober 2004 15:33 schrieb Takashi Iwai:

Quoted 7 lines

I'm afraid it's still racy.
In usX2Y_urbs_start(), the complete callback may be called between usb_submit_urb() and the loop of schedule_timeout().
So, the loop should check at first the condition before sleeping like below:

Quoted 8 lines
	while (atomic_read(&subs->state) != state_PREPARED) {
		timeout = schedule_timeout(timeout);
		if (signal_pending(current))
			...
		if (! timeout)
			...
	}
Quoted 6 lines

<click, beep, ring> ok, thanks!
But there is still a small race window between

	while (atomic_read(&subs->state) != state_PREPARED) {
and   /* right here  the complete callback may be called */
		timeout = schedule_timeout(timeout);

no?

Oh yes, sure.

Quoted 1 line

Is there a trick to avoid even this small race window?

As I wrote before, use spinlock around it or completion. The latter might be easier although it's UNINTERRUPTIBLE only.

Quoted 9 lines
Quoted 4 lines

Regarding the new config stuff:
Let's kill CONFIG_SND_USB_USX2Y_NRPACKS_VARIABLE option. No one will want to disable such a module option.

Well, me wants to disable it as it works best with nrpacks=1 here and I'd like the machine code to not contain the unneeded loops then...
(gcc optimizes those loops away if it reads "#define nrpacks 1") Can I keep some (commented out?) define switch inside the source (invisible to KConfig) to suit my needs?

Then it's ok. Having a new stuff in Kconfig would make it difficult to build for older systems.

Quoted 6 lines
Quoted 5 lines

Also, please avoid the Kconfig int value as much as possible if you want to support the driver for 2.2/2.4 kernels, too. I believe it's ok to put a hardcoded value 4 like usbaudio.c if you describe the module option properly in the document.

Erm, which document?

Create a new one for your driver ;)

Quoted 2 lines

And I'm not too keen to do the driver (sup/back)port for 2.2/2.4 kernels myself. I'll avoid the Kconfig int value, making it simpler for volunteers.

Ok, thanks.

Takashi

-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal Use IT products in your business? Tell us what you think of them. Give us Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl

#10 lore raw

Re: [PATCH]snd-usb-usx2y 0.8.1 against alsa-kernel cvshead: OHCI doesn't crash anymore

Am Mittwoch 20 Oktober 2004 11:35 schrieb Takashi Iwai:

Quoted 25 lines

At Wed, 20 Oct 2004 02:28:15 +0200,

Quoted 15 lines · Karsten Wiese

Am Dienstag 19 Oktober 2004 15:33 schrieb Takashi Iwai:

Quoted 6 lines

I'm afraid it's still racy.
In usX2Y_urbs_start(), the complete callback may be called between usb_submit_urb() and the loop of schedule_timeout().
So, the loop should check at first the condition before sleeping like below:

Quoted 7 lines
	while (atomic_read(&subs->state) != state_PREPARED) {
		timeout = schedule_timeout(timeout);
		if (signal_pending(current))
			...
		if (! timeout)
			...
	}
Quoted 6 lines

<click, beep, ring> ok, thanks!
But there is still a small race window between

	while (atomic_read(&subs->state) != state_PREPARED) {
and   /* right here  the complete callback may be called */
		timeout = schedule_timeout(timeout);

no?

Quoted 7 lines

Oh yes, sure.

Quoted 1 line

Is there a trick to avoid even this small race window?

As I wrote before, use spinlock around it or completion. The latter might be easier although it's UNINTERRUPTIBLE only.

I used the wait_event macro in this version. Its also used in snd_card_free(), so I hope stuff is not too hard to port to 2.4 and 2.2 kernels.

Quoted 20 lines
Quoted 9 lines
Quoted 3 lines

Regarding the new config stuff:
Let's kill CONFIG_SND_USB_USX2Y_NRPACKS_VARIABLE option. No one will want to disable such a module option.

Well, me wants to disable it as it works best with nrpacks=1 here and I'd like the machine code to not contain the unneeded loops then... (gcc optimizes those loops away if it reads "#define nrpacks 1") Can I keep some (commented out?) define switch inside the source (invisible to KConfig) to suit my needs?

Then it's ok. Having a new stuff in Kconfig would make it difficult to build for older systems.

Quoted 6 lines
Quoted 4 lines

Also, please avoid the Kconfig int value as much as possible if you want to support the driver for 2.2/2.4 kernels, too. I believe it's ok to put a hardcoded value 4 like usbaudio.c if you describe the module option properly in the document.

Erm, which document?

Quoted 8 lines

Create a new one for your driver ;)

Quoted 3 lines

And I'm not too keen to do the driver (sup/back)port for 2.2/2.4 kernels myself. I'll avoid the Kconfig int value, making it simpler for volunteers.

Ok, thanks.

version 0.8.5 patch is attached.

thanks,
Karsten

patch header:

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

Summary: Fixes crashes that occured when communicating over ohci usb-hcds.

Work on this started, when rumors spread that ohci equipped machines would crash.
This was due to me missing two facts:
1) Ohci has a bigger usb frame number wrap around.
2) It only supports URB_ISO_ASAP when submitting iso urbs. Fixed now.

<<<<

#11 lore raw

snd-usb-usx2y 0.8.5 against alsa-kernel cvshead: OHCI doesn't crash anymore

Quoted 12 lines · Karsten Wiese

Summary: Fixes crashes that occured when communicating over ohci usb-hcds.

Work on this started, when rumors spread that ohci equipped machines would crash.
This was due to me missing two facts:
1) Ohci has a bigger usb frame number wrap around.
2) It only supports URB_ISO_ASAP when submitting iso urbs. Fixed now.

May I ask what's the status of this, wrt.including snd-usb-usx2y version 0.8.5 on the official alsa-kernel tree?

People seem to be happy with it, specially the OHCI owners, for whom the current alsa-driver-1.0.7rc2/CVS is badly broken and causes a serious kernel oops (bug #0000425).

I hope it makes into 1.0.7 release :)

Cheers.

Signature · 11 lines
rncbc aka Rui Nuno Capela
[email protected]




-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
#12 lore raw

Re: snd-usb-usx2y 0.8.5 against alsa-kernel cvshead: OHCI doesn't crash anymore

Quoted 2 lines · Rui Nuno Capela

May I ask what's the status of this, wrt.including snd-usb-usx2y version 0.8.5 on the official alsa-kernel tree?

Could you resend this patch?

Jaroslav

-----
Jaroslav Kysela <[email protected]>

-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal Use IT products in your business? Tell us what you think of them. Give us Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl

#13 lore raw

Re: snd-usb-usx2y 0.8.5 against alsa-kernel cvshead: OHCI doesn't crash anymore

Quoted 8 lines · --- Jaroslav Kysela
Quoted 1 line · Rui Nuno Capela

May I ask what's the status of this, wrt.including

snd-usb-usx2y version

Quoted 1 line

0.8.5 on the official alsa-kernel tree?

Could you resend this patch?

wait some minutes, then I'll send the newest Version

Karsten

___________________________________________________________
Gesendet von Yahoo! Mail - Jetzt mit 100MB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de

-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal Use IT products in your business? Tell us what you think of them. Give us Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl

#14 lore raw

Re: snd-usb-usx2y 0.8.5 against alsa-kernel cvshead: OHCI doesn't crash anymore

In reply to Jaroslav Kysela
Quoted 7 lines · Jaroslav Kysela
Quoted 2 lines · Rui Nuno Capela

May I ask what's the status of this, wrt.including snd-usb-usx2y version 0.8.5 on the official alsa-kernel tree?

Could you resend this patch?

Sure. On behalf of Karten's :)

Quoted 1 line
Quoted 1 line
Quoted 1 line

Summary: Fixes crashes that occured when communicating over ohci usb-hcds.

Work on this started, when rumors spread that ohci equipped machines would crash.
This was due to me missing two facts:
1) Ohci has a bigger usb frame number wrap around.
2) It only supports URB_ISO_ASAP when submitting iso urbs. Fixed now.

<<<

#15 lore raw

Re: snd-usb-usx2y 0.8.5 against alsa-kernel cvshead: OHCI doesn't crash anymore

Am Dienstag 26 Oktober 2004 11:13 schrieb Rui Nuno Capela:

Quoted 8 lines
Quoted 5 lines · Jaroslav Kysela
Quoted 2 lines · Rui Nuno Capela

May I ask what's the status of this, wrt.including snd-usb-usx2y version 0.8.5 on the official alsa-kernel tree?

Could you resend this patch?

Sure. On behalf of Karten's :)

attached 0.8.6 has on more bug fixed ;-)

Karsten
*************
Summary: Fixes crashes that occurred when communicating over ohci usb-hcds.

Version: 0.8.6
Work on this started, when rumors spread that ohci equipped machines would crash.
This was due to me missing two facts:
1) Ohci has a bigger usb frame number wrap around.
2) It only supports URB_ISO_ASAP when submitting iso urbs. Fixed.