userland swsusp
Quoted 5 lines · Pavel Machek
This is prototype of userland swsusp. I'd like kernel parts to go in, probably for 2.6.16. Now, I'm not sure about the interface, ioctls are slightly ugly, OTOH it would be probably overkill to introduce syscalls just for this. (I'll need to add an ioctl for freeing memory in future).
I'm curious on the restrictions the userspace part would have to accept. Can /usr/swsusp.c write to a file? Currently, you allow it, but I doubt whether it would be wise to write to a file after you've snapshotted kernel's filesystem state. OTOH, I don't want to reserve a partition just for the image. Can userspace allocate memory after ioctl(SYS_FREEZE)?
I have userspace supported encryption of the image in mind.
Stefan
Hi!
Quoted 8 lines
Quoted 5 lines
This is prototype of userland swsusp. I'd like kernel parts to go in, probably for 2.6.16. Now, I'm not sure about the interface, ioctls are slightly ugly, OTOH it would be probably overkill to introduce syscalls just for this. (I'll need to add an ioctl for freeing memory in future).
I'm curious on the restrictions the userspace part would have to accept. Can
/usr/swsusp.cwrite to a file? Currently, you allow it, but I doubt
No. Writing to file would trash the filesystem. But you can bmap the file, then write to the block device.
Quoted 3 lines
whether it would be wise to write to a file after you've snapshotted kernel's filesystem state. OTOH, I don't want to reserve a partition just for the image. Can userspace allocate memory after
ioctl(SYS_FREEZE)?
Better avoid memory allocation.
Quoted 1 line
I have userspace supported encryption of the image in mind.
Yes, that should be feasible.
Pavel
Signature · 1 line
64 bytes from 195.113.31.123: icmp_seq=28 ttl=51 time=448769.1 ms
Am Mittwoch 16 November 2005 20:07 schrieb Pavel Machek:
Quoted 2 lines
No. Writing to file would trash the filesystem. But you can bmap the file, then write to the block device.
And for reading, I could used a device mapper enforced read only mount or filesystem code from grub.
Hmm, how about a possibility to ask the kernel for a list of free pages on a swap device? This way, userspace could write the image to swap as the kernel currently does, avoiding possible trouble with filesystems.
Quoted 1 line
Better avoid memory allocation.
And all memory allocated and mapped in advance would be part of the image. But this is totally acceptable for a "suspend helper".
Stefan
Hi!
Quoted 5 lines
Quoted 2 lines
No. Writing to file would trash the filesystem. But you can bmap the file, then write to the block device.
And for reading, I could used a device mapper enforced read only mount or filesystem code from grub.
Or use a filesystem that honours read-only option, like ext2...
Quoted 3 lines
Hmm, how about a possibility to ask the kernel for a list of free pages on a swap device? This way, userspace could write the image to swap as the kernel currently does, avoiding possible trouble with filesystems.
Yes, that is the plan.
Quoted 4 lines
Quoted 1 line
Better avoid memory allocation.
And all memory allocated and mapped in advance would be part of the image. But this is totally acceptable for a "suspend helper".
Yes.
Pavel
Signature · 1 line
Thanks, Sharp!
Quoted 5 lines · Pavel Machek
Quoted 2 lines
I'm curious on the restrictions the userspace part would have to accept. Can
/usr/swsusp.cwrite to a file? Currently, you allow it, but I doubtNo. Writing to file would trash the filesystem. But you can bmap the file, then write to the block device.
Do/should all the filesystems get remounted read-only as a precaution? Or is that overkill?
Rob
Hi!
Quoted 8 lines
Quoted 5 lines
Quoted 2 lines
I'm curious on the restrictions the userspace part would have to accept. Can
/usr/swsusp.cwrite to a file? Currently, you allow it, but I doubtNo. Writing to file would trash the filesystem. But you can bmap the file, then write to the block device.
Do/should all the filesystems get remounted read-only as a precaution? Or is that overkill?
You may not do that. remounting the filesystem writes to it... and that's no-no.
Pavel
Signature · 1 line
Thanks, Sharp!