bit-tech.net

Source points to Haswell Z87 USB bug

Source points to Haswell Z87 USB bug

Intel's Z87 chipset, designed for its next-generation Haswell processors, may have a bug in its USB 3.0 controller, a source has claimed.

Rumours point to a bug in Intel's upcoming Z87 chipset, designed for boards running the company's next-generation Haswell architecture processors, that cause some external hard drives to disappear during sleep mode.

According to an anonymous source with access to Z87 hardware speaking to Australian technology site PC & Tech Authority, the initial C1 stepping of the chipset has a bug in its USB 3.0 controller that causes some models of external hard drive or flash drive to enter a standby mode from which it can't be programmatically woken.

The flaw, the site's source claims, triggers when the computer enters the S3 sleep mode - a suspend-to-RAM state in which the computer draws very little power while retaining all running applications and data. As expected, all connected storage devices enter a sleep state when the computer is in S3 - but not all of them wake up again when the computer returns to a running state. Due to the bug, it is claimed, some drives will remain disconnected until they are physically removed from their USB port and plugged back in - a process which will dismount and then re-mount the drive, potentially losing any files which were open at the time the computer entered sleep mode.

It's not a serious flaw, to be sure, and a far cry from the design issue in Intel's Sandy Bridge SATA chipsets that led to multiple manufacturers halting shipments of affected motherboards until the issue could be resolved with a new stepping. It's also unlikely to halt Intel's scheduled roadmap for the Haswell launch: as such a minor issue, and affecting as it does only selected external storage devices, it's likely Intel will ship the parts to its customers anyway while working to resolve the problem with a C2 stepping as quickly as possible.

That is, if the flaw exists at all: Intel, as is usual, has refused to comment on the matter, rebuffing the site's request for clarification with a brusque statement that it 'won't comment on NDA (Non-Disclosure Agreement) technical work we are doing,' while stating that it is fully on-track for launching Haswell and its supporting motherboards.

11 Comments

Discuss in the forums Reply
Pookie 8th March 2013, 12:02 Quote
An Intel chipset with bug!!!! Blimey that's unheard of ;)
rollo 8th March 2013, 12:30 Quote
This is the reason people do testing, rumours are us ftw.
ZeDestructor 8th March 2013, 12:41 Quote
Quote:
Originally Posted by rollo
This is the reason people do testing, rumours are us ftw.

Indeed. If this is indeed true, I suspect Intel may move only the C2 (or newer revisions) to final manufacturing...
adidan 8th March 2013, 13:19 Quote
Can't see them going to market before everything's ironed out, i highly doubt another SB issue.

Possible flaw with unreleased tech. Not sure what there is to take from that really. Anonymous rumours, about as reliable as astrology.
r3loaded 8th March 2013, 13:22 Quote
Hardware logic design - it literally is far more complicated than rocket science.

Sent from my Nexus 4 using Tapatalk 2
tad2008 8th March 2013, 18:18 Quote
Well I for one hope the motherboard manufacturers stand by their guns again and refuse to buy any of the C1 chips at all, the military and businesses wouldn't settle for it why should we as consumers be forced to pay for their mistake and the hours of grief and misery it would cause not to mention the harm, however minor to the PC industry as a whole.
POMLORE 8th March 2013, 19:35 Quote
To be fair if you have an SSD its not an issue, sleep mode should be dissabled
ffjason 8th March 2013, 21:48 Quote
Is this really news?

This has been able to happen with Windows Sleep mode for a few years now - across the last 3 generations of chipset!! I recall first encountering this on X58 chipset but it was a bug before this as well. It's more likely to be a software bug rather than hardware related. Something Microsoft should have fixed years ago.

It's specifically related to USB Suspend (a windows feature) and the ability for hard drives to enter standby mode (another windows feature).

The fix is obviously to disable sleep, usb suspend and the hard drive time out in the Power Config. However, turning off the hard drive suspend & sleep mode feature in the Registry is preferred.
Gareth Halfacree 9th March 2013, 11:11 Quote
Quote:
Originally Posted by ffjason
It's specifically related to USB Suspend (a windows feature)
That's not a Windows feature, it's part of the formal USB specification and supported on all operating systems that implement USB. My Linux systems, for example, support all three USB power modes quite happily (Attached, Powered, Suspended) without running Windows.
Quote:
Originally Posted by ffjason
and the ability for hard drives to enter standby mode (another windows feature).
That's not a Windows feature either. That's part of the IDE specification and was adopted by the SATA specification as well using the same command set. The package hdparm on Linux allows fine-grained control over hard-drive power modes (Active/Idle, Standby and Sleep.)
Quote:
Originally Posted by ffjason
The fix is obviously to disable sleep, usb suspend and the hard drive time out in the Power Config. However, turning off the hard drive suspend & sleep mode feature in the Registry is preferred.
That's not a fix, that's a workaround. A fix lets you use the system as the various specifications require; a workaround makes it not crash any more. All my Linux boxes are left running 24/7, but enter S3 standby at night to save electricity. The hard drives and USB devices enter their low-power states as expected, and come back again when they wake up in the morning.
adidan 9th March 2013, 12:03 Quote
All good points G-man but has this moved beyond an anonymous rumour? I can't really see it worth debating until it has.
ffjason 10th March 2013, 11:53 Quote
@Gareth

I get your point, and I was wrong to refer to them as "Windows Features", and I'm not aware that this has ever been broken in Linux.

It's the Windows Implementation that's broken, that was what I meant.

I suppose this would be bad news for linux users. But my original point was that this feature hasn't worked in Windows for the best part of the last 5 years and possibly longer. It's not something I've ever used and not something I will ever plan to use until it works consistently within my main OS.

My experience in this comes from working for a systems integrator and having seen thousands of systems come back to us with issues relating to this (it can even cause BSOD's). Needless to say that we have disabled those features by default to prevent further problems. As far as we are concerned the problem is fixed :)
Log in

You are not logged in, please login with your forum account below. If you don't already have an account please register to start contributing.



Discuss in the forums