Re: XZ compression for the initramfs

2016-01-05 Thread Martin Michlmayr
* Roger Shimizu [2016-01-06 01:01]: > I suggest it's better to inform the following in NEWS: The note I was going to add to release-notes would be QNAP specific (not Orion) since they only have 4 MB flash for the ramdisk. But if initramfs-tools implements the "auto" option Ben mentioned, this wo

Re: XZ compression for the initramfs

2016-01-05 Thread Roger Shimizu
On Mon, Jan 4, 2016 at 3:58 AM, Martin Michlmayr wrote: > It goes from about 1 minute to 2 minutes and 20 seconds (without running > flash-kernel) which is a huge increase percentage wise but a minute more > is imho not a big deal, especially if it means avoiding troubles because > the ramdisk doe

Re: XZ compression for the initramfs

2016-01-03 Thread Martin Michlmayr
* Roger Shimizu [2016-01-04 00:54]: > Size of initrd is always an issue on embedded platform. Surely XZ can > extremely reduce the size, however as I remember, it takes MUCH more > time to finish update-initramfs command. > > Did you evaluate how long does it take to generate one initramfs image

Re: XZ compression for the initramfs

2016-01-03 Thread Lennart Sorensen
On Mon, Jan 04, 2016 at 01:21:49AM +0900, Roger Shimizu wrote: > I don't worry about memory usage because, according to xz's manpage, > it simply can be solved by adding 'export > XZ_DEFAULTS=--memlimit=64MiB"' (or 32MiB or even lower based on your > box's spec) to your ~/.bashrc. Could do the job

Re: XZ compression for the initramfs

2016-01-03 Thread Lennart Sorensen
On Sun, Jan 03, 2016 at 11:04:22AM -0500, Alan Corey wrote: > What about the RAM usage? I can't even use it often on my devio.us > account because there's shared RAM with other users. Seems like an > embedded system might have a problem with that too. Well good question. Seems for max compressi

Re: XZ compression for the initramfs

2016-01-03 Thread Roger Shimizu
On Mon, Jan 4, 2016 at 1:04 AM, Alan Corey wrote: > What about the RAM usage? I can't even use it often on my devio.us > account because there's shared RAM with other users. Seems like an > embedded system might have a problem with that too. I don't worry about memory usage because, according t

Re: XZ compression for the initramfs

2016-01-03 Thread Alan Corey
What about the RAM usage? I can't even use it often on my devio.us account because there's shared RAM with other users. Seems like an embedded system might have a problem with that too. On 1/3/16, Lennart Sorensen wrote: > On Mon, Jan 04, 2016 at 12:54:11AM +0900, Roger Shimizu wrote: >> Size o

Re: XZ compression for the initramfs

2016-01-03 Thread Lennart Sorensen
On Mon, Jan 04, 2016 at 12:54:11AM +0900, Roger Shimizu wrote: > Size of initrd is always an issue on embedded platform. Surely XZ can > extremely reduce the size, however as I remember, it takes MUCH more > time to finish update-initramfs command. > > Did you evaluate how long does it take to gen

Re: XZ compression for the initramfs

2016-01-03 Thread Roger Shimizu
> Any objections/feedback? > > I'm wondering if there are other devices with the same issue. > Unfortunately, I cannot remember if there's a ramdisk limit for the HP > mv2120 but I don't think so. The various plug devices boot from disk > and the newer QNAP devices have 9 MB space for the ramdisk.

Re: XZ compression for the initramfs

2016-01-01 Thread Martin Michlmayr
* Ian Campbell [2016-01-01 17:46]: > > How about custom kernels, though? > > In general wouldn't we say that people who want to use those are > responsible for ensuring that it meets the requirements of the rest of > the system and paying attention to changelog+NEWS etc? I agree with Ian, althou

Re: XZ compression for the initramfs

2016-01-01 Thread Ian Campbell
On Fri, 2016-01-01 at 13:03 +, Ben Hutchings wrote: > On Fri, 2016-01-01 at 12:22 +, Ian Campbell wrote: > > On Thu, 2015-12-31 at 10:50 -0800, Martin Michlmayr wrote: > > > > > > [...] > > > CONFIG_RD_XZ is enabled at least in jessie and stretch. > > > > > > Any objections/feedback? > >

Re: XZ compression for the initramfs

2016-01-01 Thread Ben Hutchings
On Fri, 2016-01-01 at 12:22 +, Ian Campbell wrote: > On Thu, 2015-12-31 at 10:50 -0800, Martin Michlmayr wrote: > > > > [...] > > CONFIG_RD_XZ is enabled at least in jessie and stretch. > > > > Any objections/feedback? > > Given the above I wonder if we should consider making XZ the default

Re: XZ compression for the initramfs

2016-01-01 Thread Ian Campbell
On Thu, 2015-12-31 at 10:50 -0800, Martin Michlmayr wrote: > > [...] > CONFIG_RD_XZ is enabled at least in jessie and stretch. > > Any objections/feedback? Given the above I wonder if we should consider making XZ the default in initramfs-tools, if CONFIG_RD_XZ was enabled in Wheezy I'd say this

Re: XZ compression for the initramfs

2015-12-31 Thread Marc Singer
On 12/31/15 10:50 AM, Martin Michlmayr wrote: I've seen some reports that the initramfs doesn't fit in flash on QNAP [deletia] 2) Create this config file by default in d-i (on these devices). Any objections/feedback? Nothing negative. This has been a recurring problem. Seems like a sm

XZ compression for the initramfs

2015-12-31 Thread Martin Michlmayr
I've seen some reports that the initramfs doesn't fit in flash on QNAP TS-109/TS-209 and TS-409 when using LVM or RAID. On my TS-109 without LVM or RAID, I have: -rw-r--r-- 1 root root 4028050 Dec 31 10:04 initrd.img-4.4.0-rc6-orion5x The flash partition is 4 MB, so this is 96%. This is already w