New Nokia N800 Firmware to fix SDHC Bug Released [Archive] - maemo.org

Tomas

10-02-2007, 07:59 AM

http://europe.nokia.com/link?cid=PLAIN_TEXT_80293

OS 2007 edition v. 4.2007.38-2 includes:

Fix to avoid SDHC memory card data corruption

This new release provides a software fix for a bug which may have appeared while copying large amount of data over-the-air or from external SDHC memory card to internal memory or vice versa. In some cases the downloaded data may have been corrupted.

There are no other changes to previous release v. 4.2007.26-8 that included Skype client support, Adobe® Flash® 9 browser plug-in an up to 8 GB memory card support among other improvements.

I guess it will take a while to hit 2007HE, but at least it's on the way now. I've had the memory card screwed up a few times by this bug.

Alvin

10-02-2007, 08:47 AM

I hope this fixes it!!

another linky
http://maemo.org/news/announcements/view/1191328183.html

Mara

10-02-2007, 08:54 AM

Works fine to me, based on few days experience. ;)

Karel Jansens

10-02-2007, 09:24 AM

I never upgraded to the Firmware-From-Hell, so will this update (if I decide to apply it, that is) bring my N800 up to date, or do I need to install the Garbage Firmware first?

Mara

10-02-2007, 09:33 AM

I never upgraded to the Firmware-From-Hell, so will this update (if I decide to apply it, that is) bring my N800 up to date, or do I need to install the Garbage Firmware first?

It is the complete firmware image. If you flash the N800 firmware using the "official" method you'll be up to a newest OS. (And, of course you need to reinstall all your apps... :eek: Making backup and restoring after flash would help to restore most system and application specific settings.)

luca

10-02-2007, 10:12 AM

is the flasher "--flash-only kernel" option supposed to work? (not that I tried and it didn't work, just wanted to make sure before I try)

Edit: don't bother replying, I just saw this post http://www.internettablettalk.com/forums/showpost.php?p=79099&postcount=61

ImDisaster

10-02-2007, 10:25 AM

It is ludicrous that you have to flash new firmware and lose all your programs, to fix this bug....especially for such a minor update with only one fix. Come on Nokia, release a small update that can be downloaded from the applications list.

zerojay

10-02-2007, 10:35 AM

Nokia already knows and that functionality is supposed to be a part of Chinook. After that, all updates will be done through application manager.

Karel Jansens

10-02-2007, 12:06 PM

It is the complete firmware image. If you flash the N800 firmware using the "official" method you'll be up to a newest OS. (And, of course you need to reinstall all your apps... :eek: Making backup and restoring after flash would help to restore most system and application specific settings.)

Thanks. I think I'll wait a couple of weeks anyway; it sure was the right way to go with the previous firmware.

And as to backup: Won't be of much help to me; I've got too many applications installed without Application Manager (which sucks anyway).

I really, really hope this is the last time we have to go through this hassle just to update the O/S. I'm so fed up with it, sometimes even the iPhone (<spit!>) starts looking good.

Karel Jansens

10-02-2007, 12:07 PM

Nokia already knows and that functionality is supposed to be a part of Chinook. After that, all updates will be done through application manager.

That doesn't sound reassuring.

geneven

10-02-2007, 12:22 PM

Yes, even I am irritated by having to install everything again. I recently had to install all my programs TWICE IN ONE DAY because of a problem I have mercifully forgotten. Now to have to do it again is a real pain. Sometimes I would say that Linux is an operating system for people with lots of extra time (and Windows is for people with lots of extra money).

TA-t3

10-02-2007, 12:27 PM

And as to backup: Won't be of much help to me; I've got too many applications installed without Application Manager (which sucks anyway).

I strongly disagree with that. It doesn't suck.
In any case, application manager is an interface to the Debian package management system, which is basically very good. The point is thus that as long as you have your repositories (catalogues in application manager speak) in order then they will be preserved by a restore after upgrade. You'll still have to re-install though, but that's pretty painless if you don't have to hunt for repositories or .install files (it took me all of fifteen minutes to reflash, restore and reinstall all my software from repos after the previous firmware upgrade).

If at all possible one should IMO install from repositories when possible, install from .deb only when that's the sole alternative. In the latter case you should then also keep them collected in a directory on a memory card, for easy re-installation.

zerojay

10-02-2007, 12:32 PM

zerojay

10-02-2007, 12:33 PM

That doesn't sound reassuring.

Why is that?

Karel Jansens

10-02-2007, 12:34 PM

I strongly disagree with that. It doesn't suck.
In any case, application manager is an interface to the Debian package management system, which is basically very good. The point is thus that as long as you have your repositories (catalogues in application manager speak) in order then they will be preserved by a restore after upgrade. You'll still have to re-install though, but that's pretty painless if you don't have to hunt for repositories or .install files (it took me all of fifteen minutes to reflash, restore and reinstall all my software from repos after the previous firmware upgrade).

If at all possible one should IMO install from repositories when possible, install from .deb only when that's the sole alternative. In the latter case you should then also keep them collected in a directory on a memory card, for easy re-installation.

Oh, Debian package management is Good <TM>. But Application Manager is crap. And your advice to install from repositories is a particularly nice example of the crappiness of Application Manager: It takes just the one repository to be offline for the wrong five minutes to have you wading through logfiles to find out where Crapplication Manager choked on.

That is so eighties software...

Karel Jansens

10-02-2007, 12:35 PM

Why is that?

See my remarks on Crapplication Manager...

zerojay

10-02-2007, 01:02 PM

Karel... I.... I just don't even have words for your explanation.

penguinbait

10-02-2007, 01:04 PM

Can someone PM me a kernel image, I have no access to pc, i am out of town, and not returning until Friday I can use fanoush's kernel flasher, but I do not have a way to unpack it???

Karel Jansens

10-02-2007, 01:44 PM

Karel... I.... I just don't even have words for your explanation.

Well... For starters, it's not an explanation, it's a rant. I'm good at those.

fpp

10-02-2007, 01:47 PM

http://europe.nokia.com/link?cid=PLAIN_TEXT_80293
I guess it will take a while to hit 2007HE, but at least it's on the way now. I've had the memory card screwed up a few times by this bug.

Why should it concern 2007HE ? The 770 doesn't use SD cards, much less SDHC...

Texrat

10-02-2007, 02:00 PM

What if he had added "new and IMPROVED! " in front of "application manager"?

:D

Karel Jansens

10-02-2007, 02:05 PM

What if he had added "new and IMPROVED! " in front of "application manager"?

:D

Oh yeah. Sure. That'll convince me.

After all, it's been crappy only... What? Two years?

zerojay

10-02-2007, 02:20 PM

mobiledivide

10-02-2007, 02:29 PM

Since I have never had any memory card issues (touch wood) I am going to wait for a more substantial upgrade to come out before upgrading unless anyone can give me a compelling reason to do it (2 *2gb sandisk cards).
To continue on the thread derailment one of the things that slightly put me off recommending linux for noobs was the command line application installation and adding libraries etc, the application manager really takes a lot of that work out of it and with a basic understanding of what you are doing it really does go far to making things "just work".ew

Mara

10-02-2007, 02:42 PM

After all, it's been crappy only... What? Two years?

Watch out... there is an improvement coming to crapplication manager in Chinook... (If there is something wrong, it will tell you what's wrong...) ;)

geneven

10-02-2007, 02:47 PM

It doesn't suck for me. How do I get it to suck?

One weird thing is that now that I installed the 'fix', I find that opening and closing programs is accompanied by a loud noise I've never heard before, and I've been using my N800 since January.

Edit: I was able to turn off the sounds in control panel, but when I ran Rhapsody, I could only get sound that was too quiet or way distorted and loud, making classical piano sound like Jimi Hendrix, sort of.

So, I decided to try headphones, and that was normal. And when I went back to the speakers, they were back to normal...

My theory is that the speakers were outputting sound assuming that headphones were plugged in, and plugging headphones in caused the N800 to come to its senses? Something like that...

penguinbait

10-02-2007, 02:57 PM

Kernel image anyone??? Can someone PM me the new kernel image please???

Nik1

10-02-2007, 03:09 PM

What do you suggest to people who are not having memory corruption errors? I only have one 4 GB sdhc card and have not experienced any problems. Is there any other problems this firmware fixes or does it just cover the SDHC bug? At this point I am very relectant to reinstall all my programs...

zerojay

10-02-2007, 03:34 PM

It also covers large download problems. I would make the update simply to protect your card from failing.

Reggie

10-02-2007, 03:45 PM

N800 firmware update time... From Maemo:Nokia has released a new version of the Internet Tablet OS 2007 edition, available now for download. This new release (version number 4.2007.38-2) provides a software fix for a bug which may have appeared while copying large amount of data over-the-air or from external SDHC memory card to internal memory or vice versa. In some cases the downloaded data may have corrupted. This new release fixes also a bug which drops in some cases voltage from memory card, though cards should manage this kind of situations without corruption.Download the firmware now.Read the full article. (http://www.internettablettalk.com/2007/10/02/new-nokia-n800-firmware-to-fix-sdhc-bug-released/)

fpp

10-02-2007, 03:46 PM

I don't see how servers being unavailable (still hasn't happened to me even once) has anything to do with the App Manager being crappy, but whatever you say, Karel. Whatever you say. Go on finding random reasons to complain.Well, you may not like the term "crappy", but I agree with Karel that the app manager is highly perfectible, both in its ergonomics and functionality.

Indeed, repositories being sometimes unavailable are not the app manager's fault (but should we need so many ?...). OTOH, it could and should be much easier to enable/disable them as needed without painstakingly editing them one by one through that tiny list window that looks like it was designed by Torquemada...

Also, you'd think that instead of the three lone, useless and disabled buttons making up the tool bar on the app's main screen, there could be shortcuts for useful functions available only from submenus : install package from file, refresh package list, show installer log, manage repos...

Apart from these ergonomics 101 basics, the app manager *does* have an unfortunate tendency to end up in a sort of limbo where it doesn't really work anymore as far as installing and uninstalling are concerned. I suspect many users just shrug it off and take advantage of the next firmware update to flash the problem away ; others use black-magic commands in xterm to straighten it up, using one after another gleaned in the forums until one works.

In either case it's not the fault of the repos, and not the user's (happens even without tinkering with Red pill mode or command line dpkg or whatever). It's just that the app is not as robust and reliable as we usually expect a package manager to be : and that is, huh... looking for the appropriate word here... well, yes, crappy.

In this sense I understand Karel's misgivings about relying on this tool to manage entire OS life cycles from Chinook onwards, like we do on desktops and servers, without the recurrent "clean sheet" firmware flashing we've had till now (no matter how much we hate 'em), and without assurance that it's been (being ?) *seriously* upgraded...

Milhouse

10-02-2007, 03:51 PM

Kernel image anyone??? Can someone PM me the new kernel image please???

Here you go, just the kernel image for 4.2007.38-2....

http://nmacleod.com/nokia/n800/zImage-4.2007.38-2 (1,317,632 bytes)

fanoush

10-02-2007, 04:04 PM

You may better also flash initfs (and wait a bit for updated bootmenu). Whole initfs is recompiled. Maybe from same source but I guess at least wlan driver modules are new and may contain some (or even most?) fixes.

penguinbait

10-02-2007, 04:11 PM

Moonshine

10-02-2007, 04:17 PM

Initfs has changed as well, how much we'll only know once fanoush has a chance to take a look. When I tried his initfs_flash utility, with a fresh new install, it exited and warned that there was a new initfs version being used.

fanoush

10-02-2007, 04:24 PM

My understanding is the only thing that changed was the kernel, is this not the case??
Kernel modules in initfs are also part of kernel. I guess fix for corruption when downloading via wlan may be in wlan driver. Modules in initfs are recompiled so this may be the reason why.

BTW, I have updated initfs_flasher.tgz (http://fanoush.wz.cz/maemo/initfs_flasher.tgz) with changes for this FW. It is untested but I hope it will work. There are no changes in linuxrc script which is the only thing that matters for bootmenu.sh script. USB recovery mode is also included but you may always answer no if you don't want it.

penguinbait

10-02-2007, 04:30 PM

My initial guess, was the ext2/3 modules, not wlan. But was just a guess...

Milhouse

10-02-2007, 04:42 PM

I just stuck the new kernel on, and it seems OK so far - booted fine, got WiFi etc.

Annoying to see that Nokia still don't output the card speed to dmesg - that's a really useful feature present in the "community" SDHC kernel... guess I'll need to submit an enhancement request for a flaming obvious feature that Nokia leave out.

Moonshine

10-02-2007, 04:42 PM

Excellent! Worked for me, thanks fanoush! Now I can get back to work. :)

(and hope and wait for Philip to make another kernel with 48MHz support...)

fanoush

10-02-2007, 04:51 PM

I just stuck the new kernel on, and it seems OK so far - booted fine, got WiFi etc.

Yes, but you may still not get the fix. On the other hand kernel modules are changed but size is still the same so maybe it is just rebuild from same source. We may know more when someone from Nokia answers bug #1204 (https://bugs.maemo.org/show_bug.cgi?id=1204)

penguinbait

10-02-2007, 05:06 PM

Yes, but you may still not get the fix. On the other hand kernel modules are changed but size is still the same so maybe it is just rebuild from same source. We may know more when someone from Nokia answers bug #1204 (https://bugs.maemo.org/show_bug.cgi?id=1204)

flashed initfs and kernel, using fanoush's great scripts,

booting from mmc, thanks thanks

now to test if it actually fixed anything...

was there any word on what actual problem was? varying power supplied to SD slots??

Milhouse

10-02-2007, 05:06 PM

Yes, but you may still not get the fix. On the other hand kernel modules are changed but size is still the same so maybe it is just rebuild from same source. We may know more when someone from Nokia answers bug #1204 (https://bugs.maemo.org/show_bug.cgi?id=1204)

Indeed, I'm quite interested to read the gory details of this problem! Until then I'll run the new kernel and see what happens... ;)

I've also raised bug #2062 (https://bugs.maemo.org/show_bug.cgi?id=2062) asking Nokia to add support to the kernel so that memory card characteristics are output to dmesg/syslog. :) It's a minor enhancement, but quite handy all the same.

penguinbait

10-02-2007, 05:09 PM

Here you go, just the kernel image for 4.2007.38-2....

http://nmacleod.com/nokia/n800/zImage-4.2007.38-2 (1,317,632 bytes)

I almost forgot!!!!

THANK YOU!!!!!

BarneyC

10-02-2007, 06:40 PM

No problems so far running latest with 8Gb Class6 SDHC Transcend card.

Only problems noted are that a few repositories aren't getting updated (incompatability taggy thing?) Red pill mode seems to bring them back to life mind.

BarneyC

10-02-2007, 06:43 PM

My bad.

the microb repository seems to be up the swannie.

Any ideas on how to get microb to install when the repos won't appear?

maxilogan

10-02-2007, 06:45 PM

And as to backup: Won't be of much help to me; I've got too many applications installed without Application Manager (which sucks anyway).

Not to mention that restoring a previous backup on Skype firmware will screw up *lots* of thing on your beloved N800 (sometimes the wallpaper won't redraw, the first time you boot up the home screen is all messed up etc.)

luca

10-02-2007, 06:49 PM

My bad.

the microb repository seems to be up the swannie.

Any ideas on how to get microb to install when the repos won't appear?

Well, I'm having problems too with the repositories, I cannot install the microb engine (well, actually I can, whan I cannot install is the eal component, anyway the microb engine even if installed and selected isn't working, I always get opera), I cannot install python, and I still see many of the system package as upgradeable (which I won't do, since I tried it before and I had to reflash since the tablet would stay with the initial nokia screen).

brianfcox

10-03-2007, 12:35 AM

Well I have installed this update twice tonight, and I am going back. All I get on the update is a cyclic shut down that requires me to remove the battery to restart the N800

fanoush

10-03-2007, 02:34 AM

We may know more when someone from Nokia answers bug #1204 (https://bugs.maemo.org/show_bug.cgi?id=1204)
We got response from Quim, flashing kernel shoul be enough so those who don't wan't to reinstall may just flash kernel (see kernel link posted by Milhouse above, if you can't use Nokia linux flasher you may flash it directly from device (http://www.internettablettalk.com/wiki/index.php/How_to_flash_the_SDHC_kernel_w/out_connecting_to_a_computer)).

qgil

10-03-2007, 02:51 AM

Hi there, wider response (including some collateral topics discussed here) at http://flors.wordpress.com/2007/10/03/sdhc-bugfix-release-updating-software-in-the-nokia-tablets/

anidel

10-03-2007, 03:03 AM

I love you Quim and I love the way we feel part of this project, an active part of it.
I've always been frustrated by being 'just' a customer, even if I am a developer.
I love the way the Maemo team is handling communications with us as developers and customers.

Thanks. To all of you guys.

TA-t3

10-03-2007, 06:47 AM

Quim:
Thanks for the link. Very good!

(Off-topic: This is my posting #1000 to Internet Tablet Talk. Never thought it would get that far. Virtual beers for everyone! :D)

wahlau

10-03-2007, 09:15 AM

(hope i did not double post..)

thanks a lot for the kernel and scripts. i tried it and so far my n800 is still kickin. I wrote the steps down here, (http://www.wahlau.org/updating_n800_kernel_the_different_way) so perhaps others can refer to it when needed.

one thing i noticed after the update is the slight improvement of response time when i start a software or so. is it just a feeling, or generally something has been improved?

thanks agin.

fanoush

10-04-2007, 08:20 AM

I wrote the steps down here, (http://www.wahlau.org/updating_n800_kernel_the_different_way) so perhaps others can refer to it when needed.
I have updated kernel flasher so it won't print that scary message about size mismatch. Now the size is properly checked (kernel has its size stored in its header). Smaller files (=incomplete downloads) are detected and flashing is aborted. Larger and already padded kernels (those extracted from full firmware images) are properly reported. I hope there is no additional bug introduced.

penguinbait

10-04-2007, 09:22 AM

I am missing something, while this seems to have fixed my problem, the description of the issue it fixes does not match my experiences. When untarring large files from flash to mmc1 or mmc2 (2gb kingston mmc mobile or 8gb class 6 kingston) 50% of the time my system would reboot, regardless if I download to flash, mmc, or copy using card reader. This indicates to me a problem, not having anything to do with wlan, and everything to do with mmc kernel support. Will we ever know what really went on, or why Nokia took so long to acknowledge this issue? As I said before, after 2 years I am sick of being the unpaid beta tester for this product!!

As the Beastie Boys say, GET IT TOGETHER!!!!

Mara

10-04-2007, 09:35 AM

That sounds like a real problem and I can confirm it happened to me too when I tried to untar XFCE files in my tablet. (Wasn't the latest kernel that time, though... but I assume nothing changed on that respect.)

Another issue I have seen and happens on both of my tablets is that I can not run Windows disk check utility on either of my SDHC cards: 4GB Sandisk and 8GB Patriot. It just stops after about 30 seconds, the N800 display backlight come on, and Windows tells that it is unable to complete the disk check. At the same time the memory card is not seen by Windows any more. Unplugging and replugging the USB cable bring the access to the card back. The issue does not happen if the cards do not have many files, but if there is thousands of MaemoMapper files then this happens 100% of the time. It is very reproducible. If I run the same disk check on these cards using external USB card reader they can complete the same process without issues, and reports no errors on the cards.

Maybe this is fixed in the next Chinook firmware... I need to test that... ;) If not, then I'll file a bug report.

zerojay

10-04-2007, 10:55 AM

...why Nokia took so long to acknowledge this issue?

Because they weren't able to reproduce it internally due to the bug reports not having enough detail. Fixing bugs is a hell of a lot easier if you have a rule to reproduce a bug. For example, if everytime you run the alarm program, the tablet crashed. Due to the type of bug this was as well as the fact that most of the bug contributors didn't include details they didn't think was necessary (not their faults since they aren't paid professionals), it took a hell of a lot more time to figure out what the issue was. The problem was a bug in the Linux kernel and we're basically the first ones that had the right conditions to discover it. Not Nokia's fault.

penguinbait

10-04-2007, 11:38 AM

Because they weren't able to reproduce it internally due to the bug reports not having enough detail. Fixing bugs is a hell of a lot easier if you have a rule to reproduce a bug. For example, if everytime you run the alarm program, the tablet crashed. Due to the type of bug this was as well as the fact that most of the bug contributors didn't include details they didn't think was necessary (not their faults since they aren't paid professionals), it took a hell of a lot more time to figure out what the issue was. The problem was a bug in the Linux kernel and we're basically the first ones that had the right conditions to discover it. Not Nokia's fault.

Do you also believe in Santa Claus? Perhaps nokia should put instructions in the documentation, so all users would know how to file bug reports? This is just crap, the problem could be reproduced at will on my and many other tablets, perhaps a little qa testing would have proven this. I guess they consider the user community their qa?

crap crap crap,

zerojay

10-04-2007, 01:25 PM

Do you also believe in Santa Claus? Perhaps nokia should put instructions in the documentation, so all users would know how to file bug reports? This is just crap, the problem could be reproduced at will on my and many other tablets, perhaps a little qa testing would have proven this. I guess they consider the user community their qa?

crap crap crap,

If you were able to reproduce this at will, why didn't you contribute to the bug report then? You could have saved a lot of people their memory cards. Again, you clearly have no idea what's going on and you have no idea what caused the bug and no idea how QA works. Big surprise.

Texrat

10-04-2007, 01:33 PM

Zero, if all some people choose to do is skim the subject, you're wasting your time and effort leading them to water... ;)

thomasdawes

10-04-2007, 01:44 PM

I'm a fan of Nokia's devices... However, not everyone has the time to peruse a forum such as this and find out ways to "fix" their tablet. Nokia's method of support and distribution for the N770/N800 seems unintuitive.

penguinbait

10-04-2007, 02:02 PM

Well, I guess because Nokia does not pay me, this is not my job, or my problem. It was well known that large tar fiiles caused a reboot 50% of the time, perhaps nokia should read ITT?

Nor should I be responsible for doing texrats job! I was working on printing from the IT, which was sadly missing from my perspective, I think I contribute plenty thank you.

But as zeroj says, I am just clueless. Say what you need to, but some simple scripts to generate high I/O on new firmware, that supports new cards would have shown this issue. But why do that when your customers can do this for you?

I expect no more from Nokia than I expect from any large manufacturer, step up!! This is not marketed as a hacker tool, its marketed as an end user device that should be easy to use. Does nokia have obligation to test this with all the cards on the market, while stress testing, damn right they do.

zerojay

10-04-2007, 02:17 PM

Zero, if all some people choose to do is skim the subject, you're wasting your time and effort leading them to water... ;)

Yeah, but I can't help trying anyways. ;)

onion_cfe

10-04-2007, 02:35 PM

Nokia and others - Clarification very much required on the fix before I flash it and buy a new large card:

My bug report gave details of SDHC card corruption that seemed to stem from just USB transfers. It didn't die for the last time during a file transfer of any kind - just audio playback.

Even now with smaller cards I still experience larger USB transfers stopping some way through, accompanied by the screen lighting up. Is this another bug I need to get involved with, and a separate issue to my SDHC card corruption?

I'm very pleased there is potentially a fix here, but I don't feel that the description of the problem that has been fixed (corruption caused by WLAN and inter-card transfers) matches some experiences logged in bug 1204. I was using USB for all transfers up to the point where my card died, after which ,ironically, I started to avoid it and used WLAN more.

All I really want is to be told confidently that I should buy another SDHC card and start to use this device as I originally wanted to. Help always appreciated!

Moonshine

10-04-2007, 02:36 PM

It was well known that large tar fiiles caused a reboot 50% of the time, perhaps nokia should read ITT?

This is not marketed as a hacker tool, its marketed as an end user device that should be easy to use. Does nokia have obligation to test this with all the cards on the market, while stress testing, damn right they do.

So it's for simple end users that untar large files ? :D

But really just a couple things:

a) I untarred 2 different large (200+ MB gzipd) tar files on two different n800's yesterday without trouble. Both are using Kingston 8gb sdhc class 6 cards. I guess I'm in the lucky camp, or it's some card specific issue.

b) Is it even possible to constuct a list of "all" the cards on the market, much less keep up with them "all" and test "all" of them under every condition? At some point you have to design to the specs for a technology and then test a resposible amount. Does anyone test with every conceivable hardware combination? Certainly not M$ :)

penguinbait

10-04-2007, 02:43 PM

zerojay

10-04-2007, 02:59 PM

What I find very curious is the fact that they are trying to say this was a wlan issue. Are they afraid if they say it was a problem with mmc support in the kernel, people will demand that Nokia replace their cards? My problems had nothing to do with wlan, yet they seem mysteriously fixed.

I guess that is just the conspiracy theorist in me!

They *did* say it was a problem with MMC support in the kernel and we got to be the unlucky ones to be the first to discover it. No, Nokia most likely won't be liable for your cards anyways... especially since the spec says that the card should be corrupted but never broken.

penguinbait

10-04-2007, 03:01 PM

So it's for simple end users that untar large files ? :D

But really just a couple things:

a) I untarred 2 different large (200+ MB gzipd) tar files on two different n800's yesterday without trouble. Both are using Kingston 8gb sdhc class 6 cards. I guess I'm in the lucky camp, or it's some card specific issue.

b) Is it even possible to constuct a list of "all" the cards on the market, much less keep up with them "all" and test "all" of them under every condition? At some point you have to design to the specs for a technology and then test a resposible amount. Does anyone test with every conceivable hardware combination? Certainly not M$ :)
Welll yes end users following simple instructions posted by wonderful people at ITT ;) untar large files

I guess you are lucky, I have the exact same card. try the same process 4 times on the old firmware, my luck was 50/50, same with 2gb kingston mmc mobile card

Well I understand what you are saying, other companies seem to do this. IBM owns lots of Oracle so they can be sure its supported on there systems, not just IBM DB2, or Informix, they test and certify EMC/Hitachi storage, even though they sell their own IBM storage. I think this is not too much to ask, I think they could afford to do the research if they wanted to.

But really this point is moot, since it was a wlan issue, not a card issue, so no amount of testing MMC/SD cards would have helped.....YEAH RIGHT!!!!

Mara

10-04-2007, 03:10 PM

dcarter

10-05-2007, 06:17 AM

FRUSTRATION IS OVER.
...linickx figured it out.

In our beloved application manager, change your maemo extras catalogue, in the "components section" from "free non-free" to just "free".

Then, go ahead and refresh.
You will then be able to download microb, python5, aisleriot, etc. etc. etc.

dcarter

Texrat

10-05-2007, 10:15 AM

Yeah, but I can't help trying anyways. ;)

You're a good man.

Texrat

10-05-2007, 10:16 AM

Nokia and others - Clarification very much required on the fix before I flash it and buy a new large card!

Are Quim and Eero's comments in Bugzilla not detailed enough?

Mara

10-05-2007, 11:11 AM

Are Quim and Eero's comments in Bugzilla not detailed enough?

Well... they are rather detailed, but may bark the wrong tree...

In my case, when my memory card corruptions happened, I did not have WiFi connection active, nor did I have tablet hooked up to PC via USB. The only thing that my the tablet was doing was in sleep (display off), while simultaneously running MaemoMapper in moving car, and connected to Bluetooth GPS receiver. MaemoMapper application constantly listened coordinates data from the GPS receiver and updated its (non-visible, due to display dimmed) display with correct map tiles. What it actually did, it READ (note, READ, not WRITE) map tiles from the memory card. I had all the necessary map tiles loaded in the memory card so MaemoMapper did not have a need to retrieve and save any of them to the card. (Earlier I did not have data plan on my phone, thus this would have been technically impossible, and the corruptions still happened.)

So, in my case the corruption clearly occurred in different circumstances than what was described to be the root causes. However, the fix was explained with good detail. It is my current belief that this fix should work in my use case as well. Also, since I never experienced the corruption when the MaemoMapper was in full screen mode (which forces it to keep the display always on), should further prove the problem being in the power save mode and in its implementation affecting memory cards.

So far, the new kernel has worked fine for me. (But I haven't tried this MaemoMapper test on the go yet...)

zerojay

10-05-2007, 11:25 AM

You're a good man.

Haha.. hope you still think so when ThoughtFix has his poll for his spare RX-44. :P

Texrat

10-05-2007, 11:48 AM

Well... they are rather detailed, but may bark the wrong tree...

In my case, when my memory card corruptions happened, I did not have WiFi connection active, nor did I have tablet hooked up to PC via USB. The only thing that my the tablet was doing was in sleep (display off), while simultaneously running MaemoMapper in moving car, and connected to Bluetooth GPS receiver. MaemoMapper application constantly listened coordinates data from the GPS receiver and updated its (non-visible, due to display dimmed) display with correct map tiles. What it actually did, it READ (note, READ, not WRITE) map tiles from the memory card. I had all the necessary map tiles loaded in the memory card so MaemoMapper did not have a need to retrieve and save any of them to the card. (Earlier I did not have data plan on my phone, thus this would have been technically impossible, and the corruptions still happened.)

So, in my case the corruption clearly occurred in different circumstances than what was described to be the root causes. However, the fix was explained with good detail. It is my current belief that this fix should work in my use case as well. Also, since I never experienced the corruption when the MaemoMapper was in full screen mode (which forces it to keep the display always on), should further prove the problem being in the power save mode and in its implementation affecting memory cards.

So far, the new kernel has worked fine for me. (But I haven't tried this MaemoMapper test on the go yet...)

I'm confused. I just read Eero's latest remarks and the focus was on the kernel, voltages and timing-- not wifi. So I don't get the continued emphasis on wlan and wifi by posters. :confused:

Mara

10-05-2007, 12:58 PM

I'm confused. I just read Eero's latest remarks and the focus was on the kernel, voltages and timing-- not wifi. So I don't get the continued emphasis on wlan and wifi by posters. :confused:

OK, maybe it was little bit different in Eero's comment, but I was reading this. (http://flors.wordpress.com/2007/10/03/sdhc-bugfix-release-updating-software-in-the-nokia-tablets/)

EDIT: In fact the details are in the link inside that link to the bug fix details... where they reference to the large downloads to the memory card over the air (wireless). So, technically it could be via bluetooth also?

However, I was not doing any download or writes to memory card when my card s corrupted.

onion_cfe

10-05-2007, 01:10 PM

Ok, lets try to move this along because it feels like we're at cross-purposes here.

When I read this explanation:

What happened:
- Once the card has acknowledged that it had received the data,
the kernel driver switched the MMC clock off
- If the device doesn't have any active clocks (display off, no USB
connected, no RD-mode, no network activity for a while, MMC clocks
off etc), the kernel power management switched the device into sleep mode
- In sleep mode MMC ports get less power

However, the card could still be doing something internally which would
draw power from the MMC port. With some cards, this seems to have damaged
them.

...I feel I am understanding what it means, and to me it implies that the issue could just as easily occur away from WLAN and inter-card transfers, and certainly USB, which tells me that this will fix my problem, However, the text released alongside the update specifically only mentions "over-the-air" and inter-card transfers. I am emphasising WLAN because the release did..

This new release provides a software fix for a bug which may have appeared while copying large amount of data over-the-air or from external SDHC memory card to internal memory or vice versa

I have my USB related card corruption problem, I have the patch and what it is said to fix, and I have a technical post which explains the cause of the fixed problem. I would like somebody more knowledgeable than myself to join the dots between these three things.

The question "is my problem fixed?" stands, and the fact that i've had two responses wondering why i'm asking and zero responses saying "yes" leads me to believe that i'm justified in continuing to ask. Please refrain from discouraging anybody that might be williing to provide an answer.

Texrat

10-05-2007, 01:29 PM

You're leaping to an invalid conclusion about my motive, onion. My point is we are best served by stripping away inconsequential aspects of the dialog and focussing on what's truly relevant. I have made no motion whatsoever to discourage dialog of any sort, and resent such accusations.

There's a profound difference between root cause (as explained by Eero) and circumstances of occurrence (the "over-the-air" comment). I think the latter was unnecessarily specific and caused confusion... probably why I subconsciously ascribed it much lower significance than the probable root cause remarks.

So hopefully we can all disabuse ourselves of the essentially-moot wifi aspect and focus on probable root cause. In that respect, I believe your question HAS been answered, but I'll reiterate: it appears that the problem has been fixed. However, this is one of those issues that only real-world testing can confirm or deny. Nokia developers seem satisfied that such testing proves the bug fixed. Your mileage *may* vary. That's the best anyone can offer at this point.

onion_cfe

10-05-2007, 02:41 PM

Fair points, and I apologise for misreading your motive. I'm perhaps a little sparse on this forum to be writing stuff like that in the direction of better established members.

I just wish I felt more confident about buying some new 8GB cards. In line with other's comments, I don't want to be the self-funded test platform for this one - not a second time. I think what that really amounts to is me now having to wait and see if the problem completely disappears.

I certainly haven't seen anybody report a newly corrupted card in the last 3 days, so that's a good sign. There could be nothing but good news here, and I really hope that's the case.

Texrat

10-05-2007, 02:44 PM

No problem. I can't prove the problem is completely resolved of course but I have a high degree of confidence that it is. So high in fact that at the nearest opportunity I will break down and purchase an SDHC card and install it on an N800 I use solely for testing. Believe me, I run that poor sucker through the mill. ;)

Mara

10-05-2007, 03:11 PM

I agree with Texrat that the problem *should* be fixed now.

However, the official "release notes" with the bugfix is at least confusing and questionable:

QUOTE:---------------------
This new release (version number 4.2007.38-2) provides a software fix for a bug which may have appeared while copying large amount of data over-the-air or from external SDHC memory card to internal memory or vice versa. In some cases the downloaded data may have corrupted.

This new release fixes also a bug which drops in some cases voltage from memory card, though cards should manage this kind of situations without corruption.
---------------------

So, the last sentence should be the actual bug that was fixed. What bug the first sentence is referring? Was there another bug that I'm not aware??? :confused:

Texrat

10-05-2007, 04:20 PM

I blame Finnglish. :p

Saturn

10-05-2007, 06:57 PM

I have updated kernel flasher so it won't print that scary message about size mismatch. Now the size is properly checked (kernel has its size stored in its header). Smaller files (=incomplete downloads) are detected and flashing is aborted. Larger and already padded kernels (those extracted from full firmware images) are properly reported. I hope there is no additional bug introduced.

Hi Fanoush,

Thanks a lot for the effort to give us the flasher.

Really appreciated.

Chris

Saturn

10-05-2007, 07:02 PM

Saturn

10-05-2007, 10:00 PM

Here you go, just the kernel image for 4.2007.38-2....

http://nmacleod.com/nokia/n800/zImage-4.2007.38-2 (1,317,632 bytes)

Hi Mlhouse,

Thanks a lot for hosting the image.

Maybe this is a little OT but do you think your point back to: http://www.internettablettalk.com/forums/showthread.php?p=58223#post58223
is still needed with this kernel? I notice that after flashing only the kernel, my setting @ 1000 remains. Should we reset that back to default?

Thanks again,
Chris

Traecer

10-06-2007, 12:52 AM

OK, so this firmware supposedly fixes the SDHC bug(s). But has anyone actually tried using a SDHC card yet? I don't want to install the new firmware until someone else has tried it out with a large capacity card. Any guinea pigs^H^H^H^H^H^H hands-on end user reports?

Moonshine

10-06-2007, 02:12 AM

Well, I'm posting this from an n800 with two Kingston 8gb sdhc card installed. (Root partition is on the internal card also.) No problems here. I built an identical system for my business partner also, and she's doing fine as well. However, I think the Kingston cards may be safer then others relation to the bug fix.

Milhouse

10-06-2007, 09:54 AM

Maybe this is a little OT but do you think your point back to: http://www.internettablettalk.com/forums/showthread.php?p=58223#post58223
is still needed with this kernel? I notice that after flashing only the kernel, my setting @ 1000 remains. Should we reset that back to default?

Thanks again,
Chris

Only the MMC/SD corruption issue is addressed in this new firmware (specifically with kernel changes and nothing to do with WiFi) so yes, I'd say the WiFi timeout tweak is still neeed when using the new firmware/kernel.

Your setting @1000 remained after you flashed only the kernel because the gconf settings are stored in your rootfs. If you had reflashed the entire firmware on to your device, your settings would have reverted to the default (200?) and you would no doubt experience WiFi connection problems again.

Texrat

10-06-2007, 10:31 AM

Milhouse your contributions are highly appreciated.

TA-t3

10-09-2007, 11:25 AM

From following the reports in bugzilla it appears that the problem was reproduced and understood at Nokia the moment a certain Michael (IIRC) posted a report where it became clear that the card corruption was happening when the N800 light dimmed a certain time window after data was written to the card. Only then it became clear what was happening. It wasn't just a question of 'write a large file to the card, damm*t!' or some such.

You needed to have the conditions set up like described above to reproduce the problem. _None_ of the steps-to-reproduce posted earlier had included the user's light timeout settings, not to mention actually describing the connection to the light dimming. In light of this (no pun intended) I find the whining and yelling from some posters quite uncalled for. Instead, virtual beers and cheers to the bugzilla poster which actually came up with the _true_ way to reproduce the problem!

Texrat

10-09-2007, 12:05 PM

Now I'm wondering if my extended timeout settings are what prevented me from experiencing the card problems... probably so.

brendan

10-09-2007, 06:40 PM

so my question is, am i (or any of the rest of us) going to get reimbursed for the card(s) burned by n800s? seems to me something should be done about that.

Texrat

10-09-2007, 06:51 PM

so my question is, am i (or any of the rest of us) going to get reimbursed for the card(s) burned by n800s? seems to me something should be done about that.

As explained before, the avenue for reimbursement lies with the card vendors.

Mara

10-09-2007, 08:14 PM

Now I'm wondering if my extended timeout settings are what prevented me from experiencing the card problems... probably so.

I recall someone testing this theory out and it did indeed help... maybe a quick seach on forum can find out... :rolleyes:

geneven

10-09-2007, 11:19 PM

Texrat

10-10-2007, 12:04 AM

"As explained before, the avenue for reimbursement lies with the card vendors."

This is too terse. You figure out why; I'll adopt your policy and just provide my conclusion with no hint of my reasoning.

It's not *my* policy. As for being too terse, that's simply because I see no need to completely reiterate what's been mentioned numerous times already, including the reasoning. No offense intended.

blacksmith_tb

10-16-2007, 12:51 AM

I'm still getting corruption and crashing after applying the update. Which is extremely annoying, as I'm about to go on vacation, and being able to have a bunch of mp3s and some video was the whole reason I bought the N800. Has anyone else had this problem?

Rebski

10-16-2007, 07:14 AM

Geneven
"As explained before, the avenue for reimbursement lies with the card vendors."

This is too terse.

see post #109
https://bugs.maemo.org/show_bug.cgi?id=1204

Does this help?

blacksmith_tb

10-16-2007, 02:03 PM

It sheds some light, but since the last post was on 10/12, I'd think that most people aren't still having problems. My symptoms are a little different after the firmware upgrade - before, individual files and some dirs became corrupt after Canola rescanned my library, but this happened quietly, and I only noticed it as albums dropped out of the list. After the flash, (and with a new 8Gb Transcend card), Canola's scanning again causes corruption, but with more serious results - the N800 locks up on the NOKIA boot screen, and after a minute or so, powers down. I just reformatted and checked the card under OS X, and it seems fine, so I'm trying again, this time with the card's lock switch on (though I'm not sure that'll save me). -T.