illusoryphoenix writes "A few weeks ago, Tom's Hardware noted a significant reduction in battery life of the Core Duo processors it tested when USB devices were inserted. Intel claimed that Microsoft had a bug in their USB drivers, while Tom's Hardware was unable to reproduce the same result for any of the other Pentium M microarchitecures. This issue has finally been publicly confirmed by Microsoft to be a USB driver problem which keeps the processor from entering advanced sleep states."
There's a driver glitch with brand new hardware!!! It's already been two weeks and they haven't fixed it yet!! What does this mean for the computing landscape? Is this Wintel's downfall? Will Apple return to their days of prominence? The implications could be enormous!
There's a driver glitch with brand new hardware!!! It's already been two weeks and they haven't fixed it yet!! What does this mean for the computing landscape? Is this Wintel's downfall? Will Apple return to their days of prominence? The implications could be enormous!
Didn't you hear??? Apple is switching to MS Windows, I heard it from a reliable source! And Linux costs just too much to run, we're all out of alternatives!
This slashdot article is actually a dupe from January; They even link the same article. Anyway, someone posted the Microsoft knowlegebase article with the registry fix in the earlier article thread [slashdot.org]
I know that you're joking, but I have to reply to the serious parts of your post.
There's a driver glitch with brand new hardware!!!
From the TFA
When a peripheral device was connected to a USB (universal serial bus) 2.0 port, the notebook's battery life plunged at a greater rate than would normally be expected from the use of a peripheral such as a mouse or storage key.
Nope, not new hardware. USB is not new. The core duos just made the problem more obvious.
It's already been two weeks and they haven't fixed it yet!! From the TFA
Microsoft published a Knowledge Base article on the subject in July 2005, but made that information available only to PC vendors and partners, a company representative said in a statement.
So, its actually been over six months and they haven't fixed it yet.
As usual, Microsoft waits for an issue to become public before bothering to fix it.
Don't you mean "From the TFA article", or maybe "from the friggin' TFA article"?:)
FYI information, this post is courtesy of Windows XP, based on NT technology, and transimtted using NIC card features to get the message posted as ASAP as possible.
just becuase I've nothing more to contribute (except that Tom's Hardware sucks)
One could argue that keeping USB2 devices plugged for long in a laptop running on battery is a kind of new scenario. The issue is also aggrevated on a dual core machine, as the need for "deep sleep" on one of the cores is actually a quite common scenario there, whereas the effects on a single-core older Pentium M is less pronounced (especially if you're using the system for anything, like playing a MP3, which will prevent your system from going to that level of power saving most of the time anyway).
There's a driver glitch with brand new hardware!!! It's already been two weeks and they haven't fixed it yet!! What does this mean for the computing landscape? Is this Wintel's downfall? Will Apple return to their days of prominence? The implications could be enormous!
In the Linux kernel Dave Jones also found [livejournal.com] a power consumption problem with USB, but it seems (to me) that the USB spec is just f*cked up - i wonder if this microsoft issue is related.
That's what code reviews are for. A good code review can often (not always) find problems before they show up at runtime. Several years ago, I was doing a code walkthrough (less rigorous than a review) for a new feature that required the brand spanking new IE 4 to work. The code was written to look for the literal string "MSIE 4" in the User-Agent header; when IE 5 came out, the code would have failed to detect it. This was QA for non-existent products. The engi
A good code review can often (not always) find problems before they show up at runtime.
In my experience, code reviews only pick up the reasonably obvious problems - your example was an obvious problem that could be spotted a mile off. Code reviews generally don't tend to pick up problems in intricate algorithms.
Infact, looking at the user agent string _at all_ is a bug, nomatter what string you're looking for. It is the reason that browsers have to fake their UA strings (IE claims to be Mozilla, Opera often claims to be IE, etc) - if you check UA strings then you have to update the site every time a new browser is released. On the other hand, presumably your UA test was to serve up some specific code needed to work around browser bugs - that makes detecting a later version of the browser and serving up the same code to be an invalid thing to do since that later version which hasn't yet been released may not have the same bugs so you're suddenly serving up workarounds that aren't needed and may potentially break.
That said, as other people pointed out, whilest MS didn't originally spot this bug (whcih may or may not be a problem with their QA procedures), they _did_ spot it over 6 months ago and didn't bother to fix it - that's the bigger problem. I wouldn't complain too much since under existing hardware this didn't affect people much - the real problem is that they also take this attitude with security bugs, and that's more worrying (only fix the bug when it has public attention... usually coz it's being exploited in the wild)
This is good news....That means its fixable with a minor software patch. Much better then having broken hardware.
Good news for whom?
I agree that it's certainly good for people unfortunate enough to use Microsoft's operating systems - they'll be able to fix a problem with a software patch rather then a hardware patch.
However, it's certainly not good news for microsoft - the small amount of trust that people have left in MS's QA processes will be lost in the news that they found this bug over six months ago,
Hopefully, Microsoft will have the patch available as part of Windows Update, maybe in an out-of-cycle update (though I doubt this since dual-core Intel CPU's are still not that common yet on notebooks).
We've spent almost the past two weeks performing non-stop battery life testing on five notebooks with up to 4 different USB devices, testing theories, trying to pinpoint exactly what causes this problem and testing Microsoft's fix. What follows is the process that we went through in our labs when faced with this strange bug.
Microsoft outlined a fix that involved modifying the registry key for USB 2.0. However, since then the company has realized that this is an impractical fix for most users, and is working on a new fix that could involve a BIOS update patch
What! Microsift to patch the BIOS Not on my notebook
Only on some systems... But this becomes useless if the old boot-floppy method gets removed as a result, what if you have another os installed? Or you need the bios update before windows will run?
Bios updates should be useable from multiple types of bootable media, cd, usb, floppy etc...
Or the bios itself should have an update function, whereby it can read a FAT or iso9660 filesystem from USB, floppy, CD or HD and open the appropriate update program... DEC's AlphaBios did this very nicely, and giving someone
Technically, I can do a live update from Windows. But Asus doesn't seem to care enough to write decent software, as it does not recognize that mine is an Asus brand.
Microsoft outlined a fix that involved modifying the registry key for USB 2.0. However, since then the company has realized that this is an impractical fix for most users, and is working on a new fix that could involve a BIOS update patch
What! Microsift to patch the BIOS Not on my notebook
Indeed. Microsoft can easily patch their own friggin' registry monstrosity.
Patching the BIOS of the machine is an outrageously bad suggestion, and a bad precedent.
How long before MS patches everyone's BIOS into oblivion or DRM hell?
Ok, I might be overboard here, but something does not feel right. MS does confirm that it is their usb driver's fault, but a possible solution would be a BIOS upgrade??? How is not a Windows update able to fix an MS driver bug?
But if it is not really a driver bug, as the BIOS statement would indicate, why on earth would MS cover for anybody?
It does not make sense! If Chewbacca lives on Endor, you must acquit!
Well, the actual transition between power states is controlled by both the OS and the BIOS. We also have the interesting fact that the actual time spent in C3 is reported incorrectly by the common way to do it in Windows. This is highly hypothetical, but it would be possible that some code checks "am I running in C3 now, if so, poll less extensively" and that the BIOS is responsible for reporting the C3 state incorrectly.
My favourite example of this was someone posting "there is NO WAY ON EARTH simply OPENING a jpeg could trigger a virus", about a month before the jpeg buffer overflow vulnerability was found.
I know, but I had pointed out that windows had had similar things with gifs before. Heck, libpng had a pretty nasty buffer overflow not so long ago. It's a sorry state of affairs, but assuming any sufficiently complicated piece of software doesn't have security flaws is almost certainly wrong.
[Starscream holds a press conference] Ummmm, yes... we were hoping no one would notice, but it's the fricking Insecticons gathering Energon for Megatron... Again. Microsoft only got involved because they own the North American rights to all acts of evil.
Inexperienced hot-pluggers often feel this way. First, and most importantly - there's absolutely nothing wrong with you - you are hot-pluggable. Don't let a bad initial experience convince you otherwise.
The realy problem here is with technique - jabbing is never recommended. You want to firmly grasp the peripheral near its end, then gently slide it into the port. Okay, try that a few times - firm grasp - good! - and gennnntly slide it in. Now withdraw the device, and gennnntly reinsert. In and out, in and out, over and over and over again. Excellent, now you're getting the hang of it.
Although it's sometimes normal to encounter resistance inserting a peripheral into a brand-new port, this friction should disappear with use. Be extra-gentle in these circumstances, and resist the urge to just jab a device into the slot. Again, slow and easy, gently sliding in and out. Yes...yessssss! Getting frustrated and randomly jabbing with your peripheral is unlikely to result in a successful connection, and can damage your peripheral unit or the slot. Overly forceful insertions have even caused the tip of a device to snap clean off - don't let this happen to you!
Deep sleep is a separate issue. It's normal not to sleep immediately after a peripheral is inserted - the unit is in active use, and sleeping would be undesirable. Wait until interactions with the device have ceased before entering sleep.
Hopefully this has cleared up some of your concerns. Remember that hot-plugging is a perfectly normal activity, one which anyone can learn to enjoy with a little practice.
Important This article contains information about how to modify the registry. Make sure to back up the registry before you modify it. Make sure that you know how to restore the registry if a problem occurs. For more information about how to back up, restore, and modify the registry, click the following article number to view the article in the Microsoft Knowledge Base: 256986 (https://premier.microsoft.com/kb/256986/ [microsoft.com] [microsoft.com]) Description of the Microsoft Windows registry
I thought it seemed suspicious that the functionality was there and easily enabled with a registry setting, but disabled by default. The feature was obviously found to be buggy by Microsoft's QA and deliberately disabled for the release. I guess they figured that it was better to let the USB suck power than to have it fail on wakeup.
Resumed from hibernation - Now entering Zombie Sleep State(tm) *deep and sinister buzzing sounds comming from the harddrive* But seriously - if there's one thing I really miss, now that i've been using 99% linux for over 1½ year, it's proper power management features. I've tried a few distros and none of them delivered 100% working power management, such as standby. I did, however, manage to get hibernation up and running, but apparently the docs on softwaresuspend aren't perfect, and I did manage to be a
It would help if hardware vendors tested their ACPI BIOSes to check that they work with more than the weird Microsoft interpreter, and played nice with proper implementations such as those in Linux and FreeBSD as well. This is predominantly why the PM features of Windows 2000 et seq. work so well "out of the box".
It tends to work better if you use APM instead of ACPI... Linux implements ACPI according to Intel's specs, while most hardware manufacturers implement it according to what works with microsoft's broken and undocumented implementation.
APM works really well on my Thinkpad T42, even hibernation to disk works perfectly (and it uses the bios, not the os, software suspend is very flakey) with one caveat, you have to change to a text or framebuffer console before you hibernate (if you hibernate from X11 it sometim
Whelp, if the overwhelming volume of ads at Tom's Hardware wasn't enough reason, now there's bad information up there as well? Does anyone remember when Tom's Hardware used to actually be a useful site that one could actually navigate effectively, and that would load in less than one hour per page? I guess it does still feel like I'm browsing it on a dialup, so they haven't changed *that* part...
Oh My God! (Score:4, Funny)
Re:Oh My God! (Score:3, Funny)
Didn't you hear??? Apple is switching to MS Windows, I heard it from a reliable source! And Linux costs just too much to run, we're all out of alternatives!
Re:Oh My God! (Score:5, Informative)
Actually it affects Pentium Ms as well, according to Anandtech.
"It's already been two weeks and they haven't fixed it yet!!"
Microsoft first identified the issue and published a Knowledgebase article July 12, 2005. That's a little more than 2 weeks.
In fact, the regedit quickfix they're recommending was also published on that date
Parent
Re:Oh My God! (Score:2)
Dupe + Fix link (Score:2)
Re:Oh My God! (Score:5, Informative)
There's a driver glitch with brand new hardware!!!
From the TFANope, not new hardware. USB is not new. The core duos just made the problem more obvious.
It's already been two weeks and they haven't fixed it yet!!
From the TFASo, its actually been over six months and they haven't fixed it yet.
As usual, Microsoft waits for an issue to become public before bothering to fix it.
Parent
Re:Oh My God! (Score:5, Funny)
Don't you mean "From the TFA article", or maybe "from the friggin' TFA article"?
FYI information, this post is courtesy of Windows XP, based on NT technology, and transimtted using NIC card features to get the message posted as ASAP as possible.
just becuase I've nothing more to contribute (except that Tom's Hardware sucks)
Parent
Re:Oh My God! (Score:2)
*heh*
AER Redudancy [antinode.org] surrounds us.
Re:Oh My God! (Score:3, Interesting)
Re:Oh My God! (Score:5, Funny)
John Dvorak, is that you?
Parent
anandtech test (Score:4, Informative)
Re:anandtech test (Score:3, Informative)
Re:anandtech test (Score:2, Interesting)
Funny. In my own independent testing, the Windows USB driver provides about a 30% gain in battery life using Linux as my baseline.
This is good news (Score:4, Insightful)
At least we know someones QA is still working.. ( and that wouldnt be microsoft in this case )
Re:This is good news (Score:3, Insightful)
Re:This is good news (Score:2)
It existed already for other Intel processors, it was made more pronounced by the new hardware.
What's the QA for your post?
Re:This is good news (Score:2)
Re:This is good news (Score:2)
That's what code reviews are for. A good code review can often (not always) find problems before they show up at runtime. Several years ago, I was doing a code walkthrough (less rigorous than a review) for a new feature that required the brand spanking new IE 4 to work. The code was written to look for the literal string "MSIE 4" in the User-Agent header; when IE 5 came out, the code would have failed to detect it. This was QA for non-existent products. The engi
Re:This is good news (Score:4, Insightful)
In my experience, code reviews only pick up the reasonably obvious problems - your example was an obvious problem that could be spotted a mile off. Code reviews generally don't tend to pick up problems in intricate algorithms.
Infact, looking at the user agent string _at all_ is a bug, nomatter what string you're looking for. It is the reason that browsers have to fake their UA strings (IE claims to be Mozilla, Opera often claims to be IE, etc) - if you check UA strings then you have to update the site every time a new browser is released. On the other hand, presumably your UA test was to serve up some specific code needed to work around browser bugs - that makes detecting a later version of the browser and serving up the same code to be an invalid thing to do since that later version which hasn't yet been released may not have the same bugs so you're suddenly serving up workarounds that aren't needed and may potentially break.
That said, as other people pointed out, whilest MS didn't originally spot this bug (whcih may or may not be a problem with their QA procedures), they _did_ spot it over 6 months ago and didn't bother to fix it - that's the bigger problem. I wouldn't complain too much since under existing hardware this didn't affect people much - the real problem is that they also take this attitude with security bugs, and that's more worrying (only fix the bug when it has public attention... usually coz it's being exploited in the wild)
Parent
Re:This is good news (Score:3, Insightful)
Good news for whom?
I agree that it's certainly good for people unfortunate enough to use Microsoft's operating systems - they'll be able to fix a problem with a software patch rather then a hardware patch.
However, it's certainly not good news for microsoft - the small amount of trust that people have left in MS's QA processes will be lost in the news that they found this bug over six months ago,
Re:This is good news (Score:2)
Tom's was wrong (Score:5, Informative)
http://www.anandtech.com/mobile/showdoc.aspx?i=26
Where's the new logo ? (Score:5, Funny)
Get with the times Slashdot.
BIOS Fix? (Score:4, Insightful)
What! Microsift to patch the BIOS
Not on my notebook
Re:BIOS Fix? (Score:2)
Re:BIOS Fix? (Score:2)
Re:BIOS Fix? (Score:2)
But this becomes useless if the old boot-floppy method gets removed as a result, what if you have another os installed? Or you need the bios update before windows will run?
Bios updates should be useable from multiple types of bootable media, cd, usb, floppy etc...
Or the bios itself should have an update function, whereby it can read a FAT or iso9660 filesystem from USB, floppy, CD or HD and open the appropriate update program... DEC's AlphaBios did this very nicely, and giving someone
Re:BIOS Fix? (Score:2)
Re:BIOS Fix? (Score:5, Insightful)
Indeed. Microsoft can easily patch their own friggin' registry monstrosity.
Patching the BIOS of the machine is an outrageously bad suggestion, and a bad precedent.
How long before MS patches everyone's BIOS into oblivion or DRM hell?
Parent
Re:BIOS Fix? (Score:2, Informative)
But if it is not really a driver bug, as the BIOS statement would indicate, why on earth would MS cover for anybody?
It does not make sense! If Chewbacca lives on Endor, you must acquit!
Re:BIOS Fix? (Score:2)
So predictable. (Score:4, Interesting)
Re:So predictable. (Score:2)
Re:So predictable. (Score:2)
Re:So predictable. (Score:2)
power sapping bug (Score:5, Funny)
[Starscream holds a press conference]
Ummmm, yes... we were hoping no one would notice, but it's the fricking Insecticons gathering Energon for Megatron... Again. Microsoft only got involved because they own the North American rights to all acts of evil.
It's not just Microsoft... (Score:5, Funny)
Re:It's not just Microsoft... (Score:5, Funny)
The realy problem here is with technique - jabbing is never recommended. You want to firmly grasp the peripheral near its end, then gently slide it into the port. Okay, try that a few times - firm grasp - good! - and gennnntly slide it in. Now withdraw the device, and gennnntly reinsert. In and out, in and out, over and over and over again. Excellent, now you're getting the hang of it.
Although it's sometimes normal to encounter resistance inserting a peripheral into a brand-new port, this friction should disappear with use. Be extra-gentle in these circumstances, and resist the urge to just jab a device into the slot. Again, slow and easy, gently sliding in and out. Yes...yessssss! Getting frustrated and randomly jabbing with your peripheral is unlikely to result in a successful connection, and can damage your peripheral unit or the slot. Overly forceful insertions have even caused the tip of a device to snap clean off - don't let this happen to you!
Deep sleep is a separate issue. It's normal not to sleep immediately after a peripheral is inserted - the unit is in active use, and sleeping would be undesirable. Wait until interactions with the device have ceased before entering sleep.
Hopefully this has cleared up some of your concerns. Remember that hot-plugging is a perfectly normal activity, one which anyone can learn to enjoy with a little practice.
Parent
All a bunch of whiners.... (Score:3, Insightful)
Here's the fix... (Score:2, Informative)
Make sure to back up the registry before you modify it. Make sure that you know how to restore the registry if a problem occurs. For more information about how to back up, restore, and modify the registry, click the following article number to view the article in the Microsoft Knowledge Base:
256986 (https://premier.microsoft.com/kb/256986/ [microsoft.com] [microsoft.com]) Description of the Microsoft Windows registry
SYMPTOMS
Consider the following
Re:Here's the fix... (Score:4, Informative)
Parent
Re:Here's the fix... (Score:2)
I thought it seemed suspicious that the functionality was there and easily enabled with a registry setting, but disabled by default. The feature was obviously found to be buggy by Microsoft's QA and deliberately disabled for the release. I guess they figured that it was better to let the USB suck power than to have it fail on wakeup.
Special sleep states (Score:2, Insightful)
*deep and sinister buzzing sounds comming from the harddrive*
But seriously - if there's one thing I really miss, now that i've been using 99% linux for over 1½ year, it's proper power management features. I've tried a few distros and none of them delivered 100% working power management, such as standby.
I did, however, manage to get hibernation up and running, but apparently the docs on softwaresuspend aren't perfect, and I did manage to be a
Re:Special sleep states (Score:2)
Re:Special sleep states (Score:2)
Linux implements ACPI according to Intel's specs, while most hardware manufacturers implement it according to what works with microsoft's broken and undocumented implementation.
APM works really well on my Thinkpad T42, even hibernation to disk works perfectly (and it uses the bios, not the os, software suspend is very flakey) with one caveat, you have to change to a text or framebuffer console before you hibernate (if you hibernate from X11 it sometim
Tom's hardware, not what it used to be (Score:2)
Sorry, have to do it... can't resist... (Score:2)
Re:It's friggin WINDOWS running on friggin INTEL! (Score:2)
*shyly raises hand*
Re:Perish the thought... (Score:3, Funny)
Re:So... (Score:2)