Skip to contentSkip to conversation

annoying frequent overcurrent messages.

DISCUSSION Alan Stern 7 participants 14 messages Jul 12, 2006 – Aug 8, 2006 1 message not in the archive

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

Alan Stern

author
#1 lore raw
Quoted 6 lines · Dave Jones

I have a box that's having its dmesg flooded with..

hub 1-0:1.0: over-current change on port 1
hub 1-0:1.0: over-current change on port 2
hub 1-0:1.0: over-current change on port 1
hub 1-0:1.0: over-current change on port 2

...

Quoted 3 lines

over and over again..
The thing is, this box doesn't even have any USB devices connected to it, so there's absolutely nothing I can do to remedy this.

Well, overcurrent is a potentially dangerous situation. That's why it gets reported with dev_err priority.

Evidently it's a hardware fault. Perhaps the overcurrent-detect input lines are floating and constantly triggering as a result. It's not even clear that attaching a USB device would make the problem go away.

Since you're not using the UHCI controller on that computer, you could simply rmmod uhci-hcd (or modify /etc/modprobe.conf to prevent it from being loaded in the first place). That would stop the constant interrupts and the syslog spamming.

But as for fixing the underlying hardware problem, I don't think there's anything we can do.

Alan Stern

#2 lore raw
Quoted 21 lines · Alan Stern
Quoted 6 lines · Dave Jones

I have a box that's having its dmesg flooded with..

hub 1-0:1.0: over-current change on port 1
hub 1-0:1.0: over-current change on port 2
hub 1-0:1.0: over-current change on port 1
hub 1-0:1.0: over-current change on port 2

...

Quoted 3 lines

over and over again..
The thing is, this box doesn't even have any USB devices connected to it, so there's absolutely nothing I can do to remedy this.

Well, overcurrent is a potentially dangerous situation. That's why it gets reported with dev_err priority.

Evidently it's a hardware fault. Perhaps the overcurrent-detect input lines are floating and constantly triggering as a result. It's not even clear that attaching a USB device would make the problem go away.

Quoted 7 lines

Since you're not using the UHCI controller on that computer, you could simply rmmod uhci-hcd (or modify /etc/modprobe.conf to prevent it from being loaded in the first place). That would stop the constant interrupts and the syslog spamming.

But as for fixing the underlying hardware problem, I don't think there's anything we can do.

we could at least rate-limit the messages.

Dave

Signature · 1 line
http://www.codemonkey.org.uk
#3 lore raw
In reply to Alan Stern
Quoted 9 lines · Alan Stern
Quoted 6 lines · Dave Jones

I have a box that's having its dmesg flooded with..

hub 1-0:1.0: over-current change on port 1
hub 1-0:1.0: over-current change on port 2
hub 1-0:1.0: over-current change on port 1
hub 1-0:1.0: over-current change on port 2

...

Quoted 8 lines
Quoted 3 lines

over and over again..
The thing is, this box doesn't even have any USB devices connected to it, so there's absolutely nothing I can do to remedy this.

Since you're not using the UHCI controller on that computer, you could simply rmmod uhci-hcd (or modify /etc/modprobe.conf to prevent it from being loaded in the first place). That would stop the constant interrupts and the syslog spamming.

For the syslog spamming, you could jus emit the message once when the state is first noticed, then emit a everything good message when it clears up. There's no need to log it repeatedly during the problem period.

Ray

author
#4 lore raw
Quoted 19 lines · Ray Lee
Quoted 9 lines · Alan Stern
Quoted 6 lines · Dave Jones

I have a box that's having its dmesg flooded with..

hub 1-0:1.0: over-current change on port 1
hub 1-0:1.0: over-current change on port 2
hub 1-0:1.0: over-current change on port 1
hub 1-0:1.0: over-current change on port 2

...

Quoted 8 lines
Quoted 3 lines

over and over again..
The thing is, this box doesn't even have any USB devices connected to it, so there's absolutely nothing I can do to remedy this.

Since you're not using the UHCI controller on that computer, you could simply rmmod uhci-hcd (or modify /etc/modprobe.conf to prevent it from being loaded in the first place). That would stop the constant interrupts and the syslog spamming.

Quoted 4 lines

For the syslog spamming, you could jus emit the message once when the state is first noticed, then emit a everything good message when it clears up. There's no need to log it repeatedly during the problem period.

That's almost exactly how the driver behaves currently -- the message is printed just once when the state is first noticed. Nothing is printed when the state is cleared, and nothing gets printed repeatedly during the problem period. But then the problem recurs very quickly.

Quoted 1 line · Dave Jones

we could at least rate-limit the messages.

That's true for every message in the kernel. How do you decide which messages to rate-limit?

Note that this particular message will cause problems only in the presence of defective hardware.

Alan Stern

#5 lore raw
Quoted 26 lines · Alan Stern
Quoted 19 lines · Ray Lee
Quoted 9 lines · Alan Stern
Quoted 6 lines · Dave Jones

I have a box that's having its dmesg flooded with..

hub 1-0:1.0: over-current change on port 1
hub 1-0:1.0: over-current change on port 2
hub 1-0:1.0: over-current change on port 1
hub 1-0:1.0: over-current change on port 2

...

Quoted 8 lines
Quoted 3 lines

over and over again..
The thing is, this box doesn't even have any USB devices connected to it, so there's absolutely nothing I can do to remedy this.

Since you're not using the UHCI controller on that computer, you could simply rmmod uhci-hcd (or modify /etc/modprobe.conf to prevent it from being loaded in the first place). That would stop the constant interrupts and the syslog spamming.

Quoted 4 lines

For the syslog spamming, you could jus emit the message once when the state is first noticed, then emit a everything good message when it clears up. There's no need to log it repeatedly during the problem period.

Quoted 4 lines

That's almost exactly how the driver behaves currently -- the message is printed just once when the state is first noticed. Nothing is printed when the state is cleared, and nothing gets printed repeatedly during the problem period. But then the problem recurs very quickly.

Then the logging of the 'all cleared up' message would be better if it had a bit of hysteresis to it -- the good state is noticed, but don't log it as such until it hangs out there for a while and has had a chance to quiesce.

Ray

author
#6 lore raw
Quoted 10 lines · Ray Lee
Quoted 9 lines
Quoted 4 lines

For the syslog spamming, you could jus emit the message once when the state is first noticed, then emit a everything good message when it clears up. There's no need to log it repeatedly during the problem period.

That's almost exactly how the driver behaves currently -- the message is printed just once when the state is first noticed. Nothing is printed when the state is cleared, and nothing gets printed repeatedly during the problem period. But then the problem recurs very quickly.

Quoted 4 lines

Then the logging of the 'all cleared up' message would be better if it had a bit of hysteresis to it -- the good state is noticed, but don't log it as such until it hangs out there for a while and has had a chance to quiesce.

You didn't read what I wrote -- there _is_ no "all cleared up" message.

Alan Stern

#7 lore raw
Quoted 6 lines · Alan Stern
Quoted 4 lines

Then the logging of the 'all cleared up' message would be better if it had a bit of hysteresis to it -- the good state is noticed, but don't log it as such until it hangs out there for a while and has had a chance to quiesce.

You didn't read what I wrote -- there _is_ no "all cleared up" message.

You're right, I didn't -- my apologies. However:

Quoted 4 lines
Quoted 4 lines

That's almost exactly how the driver behaves currently -- the message is printed just once when the state is first noticed. Nothing is printed when the state is cleared, and nothing gets printed repeatedly during the problem period. But then the problem recurs very quickly.

If you change my wording from "all cleared up message" to "clearing the internal state that keeps track of whether we're still in an overcurrent situation" then my suggestion still makes sense. In short, don't believe we're out of the overcurrent state until a bit of time passes.

Before we go any further, let's make sure my suggestion could even fix anything for Dave. Dave? How often are these messages printed? Do they come in clumps? Or are they periodic? In other words, would a bit of hystersis in the state clearing code remove most of the messages for you? Or am I just stirring up trouble?

Ray

#8 lore raw
In reply to Alan Stern
Quoted 6 lines · Alan Stern
Quoted 1 line · Dave Jones

we could at least rate-limit the messages.

That's true for every message in the kernel. How do you decide which messages to rate-limit?

anything the user doesn't have any means of fixing should be able to be ignored. With dmesg filled with these, it's hard to ignore them.

Quoted 2 lines

Note that this particular message will cause problems only in the presence of defective hardware.

Defective it may be, I'm not arguing that. But spewing hundreds of
these an hour isn't going to make the user fix the problem (if they even can) any faster.

*grumbles and goes to edit modprobe.conf*

Dave

Signature · 1 line
http://www.codemonkey.org.uk
author
#9 lore raw
Quoted 10 lines · Dave Jones
Quoted 6 lines · Alan Stern
Quoted 1 line · Dave Jones

we could at least rate-limit the messages.

That's true for every message in the kernel. How do you decide which messages to rate-limit?

anything the user doesn't have any means of fixing should be able to be ignored. With dmesg filled with these, it's hard to ignore them.

That wasn't a rhetorical question. A large percentage of the messages in the USB hub driver have the potential to spam the syslog, given the right sort of hardware disfunction. But it doesn't seem practical to rate-limit all of them.

Alan Stern

#10 lore raw
In reply to Dave Jones

Ar Mer, 2006-07-12 am 13:51 -0400, ysgrifennodd Dave Jones:

Quoted 5 lines

Defective it may be, I'm not arguing that. But spewing hundreds of
these an hour isn't going to make the user fix the problem (if they even can) any faster.

*grumbles and goes to edit modprobe.conf*

dmidecode
pci svid/subdevice

submit patch

I agree entirely with the other Alan here, overcurrent is a serious item with a potential severity not that far below "your cpu fan seems to have stopped going round [Ok]"

If its a dodgy board it should be possible to id and blacklist its UHCI controller

#11 lore raw
In reply to Alan Stern

Hi!

Quoted 6 lines
Quoted 6 lines

I have a box that's having its dmesg flooded with..

hub 1-0:1.0: over-current change on port 1
hub 1-0:1.0: over-current change on port 2
hub 1-0:1.0: over-current change on port 1
hub 1-0:1.0: over-current change on port 2

Quoted 8 lines

...

Quoted 3 lines

over and over again..
The thing is, this box doesn't even have any USB devices connected to it, so there's absolutely nothing I can do to remedy this.

Well, overcurrent is a potentially dangerous situation. That's why it gets reported with dev_err priority.

Well, I see overcurrents all the time while doing suspend/resume...

Why is it dangerous? USB should survive plugging something that connects +5V and ground. It may turn your machine off, but that should be it...?
Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#12 lore raw
Quoted 14 lines · Pavel Machek

Hi!

Quoted 12 lines
Quoted 6 lines

I have a box that's having its dmesg flooded with..

hub 1-0:1.0: over-current change on port 1
hub 1-0:1.0: over-current change on port 2
hub 1-0:1.0: over-current change on port 1
hub 1-0:1.0: over-current change on port 2

...

Quoted 3 lines

over and over again..
The thing is, this box doesn't even have any USB devices connected to it, so there's absolutely nothing I can do to remedy this.

Quoted 8 lines
Quoted 2 lines

Well, overcurrent is a potentially dangerous situation. That's why it gets reported with dev_err priority.

Well, I see overcurrents all the time while doing suspend/resume...

Why is it dangerous? USB should survive plugging something that connects +5V and ground. It may turn your machine off, but that should be it...?

it's fun if your main storage resides near it on the same hub... like your suspend device
(now ok suspend-to-usb-disk is silly I suppose, but I can think of some really cool usage models that that allows in an office-less office)

author
#13 lore raw
In reply to Pavel Machek
Quoted 8 lines · Pavel Machek
Quoted 2 lines

Well, overcurrent is a potentially dangerous situation. That's why it gets reported with dev_err priority.

Well, I see overcurrents all the time while doing suspend/resume...

Why is it dangerous? USB should survive plugging something that connects +5V and ground. It may turn your machine off, but that should be it...?

The key words here are "potentially", "should", and "may".

BTW, what sort of overcurrents do you see during suspend/resume?

Alan Stern

#14 lore raw

Hi!

Quoted 5 lines
Quoted 5 lines
Quoted 2 lines

Well, overcurrent is a potentially dangerous situation. That's why it gets reported with dev_err priority.

Well, I see overcurrents all the time while doing suspend/resume...

Quoted 7 lines
Quoted 3 lines

Why is it dangerous? USB should survive plugging something that connects +5V and ground. It may turn your machine off, but that should be it...?

The key words here are "potentially", "should", and "may".

BTW, what sort of overcurrents do you see during suspend/resume?

Okay, that was on X32, and I'm not using it for now, sorry about that. "Overcurrent on port XY" was the message, IIRC, but...
Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html