[hardy] kernel should have usb persist mode built in

[GiuseppeVerde]

Binary package hint: linux-image-2.6.24-10-generic

USB persist mode (described at http://www.mjmwired.net/kernel/Documentation/usb/persist.txt) is highly desirable for some segments of the userbase (most immediately the eee users, who have only 2 or 4GB of storage but have a full USB SDcard reader built in, allowing up to 16GB more).

As the file linked to above describes, it is a config option which needs to be enabled at compile time. However, the persist mode will still not be active for any devices unless specifically enabled via the /sys interface. This will prevent any negative impact on the unsuspecting user if persist mode is compiled in to the Ubuntu kernel, while still allowing those who need it (e.g. eee users) to enable it where they need to.

EEE users must currently build a custom kernel or stop all use of the SD card and unmount it before suspending or else filesystem damage occurs.

Because it should harm nobody to enable it and significantly help a segment of the userbase, I propose enabling usb persist mode in the Hardy kernels.

[GiuseppeVerde (launchpad-digitasaru)]

Attached a file showing the pain that is mine whenever I suspend and resume (RAM). The 16G card is always present in the SD slot, which is a USB device.

Please let me know if you need anything else.

[GiuseppeVerde (launchpad-digitasaru)]

Thanks a ton, Amit. Any idea when this will hit the servers for download?

[Amit Kucheria (amitk)]

Early next week definitely. Things are being built right now.

[Matteo Collina (matteo-collina)]

Any news? Is this feature already enabled in the kernel?
How can I enable it on my eee?

[Duncan Mak (duncanmak)]

I installed hardy on my eee tonight, but suspend doesn't work.

[Amit Kucheria (amitk)]

While working with Classmate (toward the very end of Hardy), it was discovered that simply enabling USB_PERSIST is not enough. We need another patch[1] to get it to work correctly. However that patch doesn't quite work due to another patch[2] in the Hardy tree. Since getting this fixed in the short time was not possible, the USB Persist patch[1] did not make it into the Hardy release.

It is now being made available through the classmate PPA[3]. Adding that to your /etc/apt/sources.list should get you the new kernel when it is built.

Also, I am currently not sure if eeePC will also require the CONFIG_MMC_UNSAFE_RESUME option set, but I suspect it will.

[1]. http://thread.gmane.org/gmane.linux.usb.general/1899
[2]. UBUNTU: USB: EHCI: add separate IAA watchdog timer (978a8bed296d7f5d76c57018417bee79046354a1)
[3]. https://edge.launchpad.net/~cmpc-developers/+archive

[Amit Kucheria (amitk)]

The MMC unsafe resume bug is: https://bugs.launchpad.net/moblin-kernel/+bug/193177

[Amit Kucheria (amitk)]

and the MMC resume bug id: https://bugs.launchpad.net/moblin-kernel/+bug/193177

[Max Goodhart (chromakode)]

Hi Amit,

Thank you for your work providing kernels for this purpose. I've installed your -cmpc kernel packages on my Eee in hopes of regaining suspend with /home on an SD card. Unfortunately, strangely, I still get the symptomatic read issues upon resume. This is particularly strange because I have checked the config for the working kernel I have been using [1], and CONFIG_MMC_UNSAFE_RESUME is unset. I also checked the various /sys/bus/usb/devices/.../power/persist to verify that under your kernel the persist state is 1. Are there any other steps I could take to hopefully resolve this issue? Is there a version with MMC_UNSAFE_RESUME that I could try? Thank you.

[1] http://forum.eeeuser.com/viewtopic.php?id=6585&p=1

[Amit Kucheria (amitk)]

The classmate kernel now provides working USB resume support. Closing this bug.

[Matteo Collina (matteo-collina)]

The last kernel available in the classmate ppa is the -17. The -19 failed building for the 386 achitecture. Can you update the classmate ppa?