Pages: [1] 2 :: one page |
|
Author |
Thread Statistics | Show CCP posts - 0 post(s) |
Whitehound
|
Posted - 2009.05.27 15:02:00 -
[1]
Edited by: Whitehound on 27/05/2009 15:02:20 I have decided to get me an SSD and to install Linux on it. If anyone has done the same, please, let me know how it is going for you.
The drive I ordered is an Intel X25-M 80GB SSD and I plan to use XFS on it.
After looking into various file systems did I first learn some bad news. Hans Reiser did kill his wife, he confessed it and the police then dug up the remains of her. ... No more ReiserFS ever again!!!
The reason why I want to use XFS is because it allows to be optimized for strip sizes and I intend to configure it so that it supports the internal geometry of the SSD. Ext4 seems to have this ability, too, but I do not know if I want all the backward compatibility of the ext file systems.
Further, do I plan to leave /var and /home on the hard disk, but I am not sure if I should not create a swap partition on the SSD. My Linux is hardly swapping with the 4GB of memory that I have installed.
I also want to disable the atime option of the file system, but have no idea if this will get my Debian into any trouble. The deadline scheduler then seems to be the best choice, so I have read.
Is someone able to help me with my worries or knows some nice optimization tricks? -- If there is anything in your life you fear of losing, then keep your mouth shut once in a while. |
Xianthar
STK Scientific The Initiative.
|
Posted - 2009.05.27 18:41:00 -
[2]
i have a eeePC and an asus laptop with one of the first gen MLC OCZ drives, both ubuntu 9.04
you sound like you pretty much have it squared away, the X25's have pretty good write performance anyway.
i use deadline, small swap on the laptop (4gb ram), normalish sized swap on the 1gb eeepc,
I use ext4 on both i have no idea why you consider backward compatibility a bad thing and keep in mind its not truly backward compatible, you can mount an ext3 partition as ext4 but the second you turn most of the new ext4 features, you can't mount it as ext3 anymore, therefore your fresh ext4 drive will not be backward compatible with ext3...
outside of that, ext4 is simply faster than XFS at almost everything:
http://www.phoronix.com/scan.php?page=article&item=ext4_benchmarks&num=1
/home is on a separate drive on the eeepc (its got a 4gb drive and a 16gb drive)
on the laptop /home is on the same drive but is a separate partition for easy distribution changes if i feel like it
on the eeepc i have limited logging and moved firefox's cache to a 100MB ramdisk as the write speeds of the SSD's in the eeepc suck, badly, i never use the 1gb of ram on such a tiny machine anyway, not even with compiz writing fire on the screen
i have noatime set on both as well
with the Intel X25-M i wouldn't worry much, it should kick the crap out of whatever your replacing on both read and write unless you have a 17" Oprah laptop with RAID0 velociraptors
only thing i would recommend is checking for firmware updates for the drive, i know Intel (and others) are still tweaking their wear leveling and write algorithms.
i've had both for...almost a year now i guess, both were transitioned to ext4 without a problem, i ran a little script that forces a move of all the files on the drive when i did this, it forces the new features of ext4(extents and such) to be "inacted" on the existing files. ext4 in theory supports an online defrag tool which would have made such a script unnecessary but the ubuntu kernel in 9.04 doesn't have the proper patches to support this and i'm way too lazy to custom compile a kernel just for that.
both boot in like 12-15 seconds or so i guess, eeepc is usually in standby and comes out of that in like 4 seconds ish? i dun know by the time i push the power button and get my hands in position to type its ready for my password. asus takes 10-15 to come out of standby and thus i just power it down and use session memory to save power. Pretty sure the standby slowness is the fault of nvidia's binary driver, its had some issues with power management in the past.
cheers
|
Whitehound
|
Posted - 2009.05.27 19:42:00 -
[3]
Originally by: Xianthar I use ext4 on both i have no idea why you consider backward compatibility a bad thing and keep in mind its not truly backward compatible, you can mount an ext3 partition as ext4 but the second you turn most of the new ext4 features, you can't mount it as ext3 anymore, therefore your fresh ext4 drive will not be backward compatible with ext3...
I do not need backwards compatibility and I am worried that it might introduced unnecessary cargo. I have looked into some benchmarks now after you wrote about ext4. If I can have ext4 without any a major hassle then it is going to be ext4 with extents now. Only when it does not require some extra work, because of it being rather new, will I use XFS and switch to it at some later time. Yeah, but thanks for the info. I much appreciate it! -- If there is anything in your life you fear of losing, then keep your mouth shut once in a while. |
Xianthar
STK Scientific The Initiative.
|
Posted - 2009.05.28 04:03:00 -
[4]
there is no cargo, ext4 doesn't have a "ext3 compatibility mode" it still uses the same rough layout, ext4 builds on top of ext3 and was originally proposed as series of patches to ext3, it was decided to fork it off as ext4 for stability testing, a new block allocation algorithm, pre allocation, 64bits in the b-trees allowing more directories, better journaling etc. very few features in ext4 actually prevent mounting an ext4 file system as ext3 such as extents (one of the biggest features) and needing more than 32k directories. ext4 isn't some brand new file system that happens to work with ext3 but rather is a logical progression of the features available in ext3.
as for the defrag, honestly i've yet to see any conclusive evidence here we probably won't until btrfs or some other file system specifically designed with SSD's in mind is ready.
you have to compare the block allocation algorithm in the file system against the block allocation algorithm in the SSD to determine what the actual resulting block layout is in the chips, and if its optimal or not. i've yet to see a SSD manufacturer publish their algorithm and given that the write/wear algorithms seem to be changing with each manufacturers firmware update and that it's their secret sauce, i don't see this happening any time soon.
If i were the SSD manufacturers i would be optimizing my allocation algorithm against what a file system expects, that is, a revolving disk, since most file systems attempt to keep files either continuous or near each other on a rotating disk. obviously these ideas don't mean much to an SSD (wtf is a cylinder in SSD land?) but if the file system is intentionally arranging blocks in a non continuous manor because it thinks they will be near each other on disk...the SSD needs to account for that or it'll be all over the map with writes.
I am also very weary of the defrag subject in general because of the "ext3 partitions never need defrag" camp, because its a lie, its VERY dependent on work load, for example i just copied a 9MB jpg image to the root drive of this system (133GB raptor, ext3 with 21% of the disk full)
*******@***:~/Desktop$ sudo filefrag -v nikon_d3x_05b.jpg [sudo] password for **********: Checking nikon_d3x_05b.jpg Filesystem type is: ef53 Filesystem cylinder groups is approximately 1064 Blocksize of file nikon_d3x_05b.jpg is 4096 File size of nikon_d3x_05b.jpg is 9530399 (2327 blocks) First block: 20936650 Last block: 20967747 Discontinuity: Block 1 is at 20936700 (was 20936650) Discontinuity: Block 2 is at 20937720 (was 20936700) //chopped for brevity Discontinuity: Block 2290 is at 20966044 (was 20966042) Discontinuity: Block 2291 is at 20967712 (was 20966044) nikon_d3x_05b.jpg: 86 extents found, perfection would be 1 extent
that drive has been trashed mostly by growing files without preallocation, older bit torrent clients, video encoding where the final size is not known, generally anything with a continuously growing file write pattern will destroy the fragmentation on an ext3 disk. theres nothing the internal layout of an SSD can do to solve such issues as, at that point, the SSD is accepting block write requests, it doesn't have a clue they are all the same file.
the last 2 Intel X-25 firmware update have dealt almost exclusively with fragmentation issues. if i were guess we'll see file systems soon supporting defrag programs aimed directly at the SSD write case, once they stabilize their firmware.
|
Ravow
Minmatar Cosmic Encounter
|
Posted - 2009.05.28 13:11:00 -
[5]
Hi!
I have a cheapo SSD 128GB Patriot Warp v2. I read a lot of stuff for it and so... it's not very complicated and you can use any file system. BUT you need to align the partition to your write block before (512k aligned if you want to be sure).
You can do the alignement job with fdisk in block mode OR you can type:
fdisk -H 224 -S 56 /dev/sdb and use fdisk normally. The first partition will be 4k aligned (put you /boot here) and the other will be 128k aligned.
For your file system, I don't recommend you XFS because this fs is designed for large file on a server with a UPS.
EXT3/4 is good, Reiser4 with lzo compression is best (you get more space for your price). Reiserfs/reiser4 is not dead, some people like Edward and me still work for it and I can confirm you that this FS -is- stable now and faster than EXT3/4 except when you mount it.
Oh, for your fstab, use the noatime option for each partition on your SSD so you will not need to write access date to each file you access...
PS : If you choose crypted/compressed Reiser4, USE THE REISER4PROGS 1.0.7+. I found a nasty bug in fsck.reiser4 1.0.6 and it's all fixed in 1.0.7.
PS2 : Tell me if you need reiser4 kernel patch for 2.6.30.
|
Whitehound
|
Posted - 2009.05.31 15:56:00 -
[6]
Edited by: Whitehound on 31/05/2009 16:04:19 I had quite some trouble with grub in combination with ext4 and xfs, and so I dropped the idea of using these file systems. Support is claimed to exist, but it has given me nothing but trouble. I am using ext3 for now. ReiserFS is not an option.
The SSD did not deliver the promised speed at first, but only gave a 120-180 MB/s for reading. I found the right tweaks as it seems and now it runs at full speed with 240 MB/s. If anyone is interested in these tweaks then checkout my /etc/init.d-script:
# Set the I/O scheduler for each block device
if [ -d /sys/block ]; then for block_device in /sys/block/*; do if [ -d $block_device/queue ]; then model="$(cat $block_device/device/model)" case "$model" in "INTEL SSD"*) echo "Found $model, using the deadline I/O scheduler."
# Increase the total length of the queue (default: 128 blocks). echo 2048 > $block_device/queue/nr_requests
# Set the deadline I/O scheduler. echo deadline > $block_device/queue/scheduler
# Increase the number of blocks in a batch (default: 16 blocks). echo 1024 > $block_device/queue/iosched/fifo_batch
# Increase the delay for read requests (default: 500ms). echo 1000 > $block_device/queue/iosched/read_expire
# Work off read requests 4-times more often (default: 2). echo 4 > $block_device/queue/iosched/writes_starved ;; "WDC WD"*) echo "Found $model, using the anticipatory I/O scheduler."
# Set the anticipatory I/O scheduler. echo anticipatory > $block_device/queue/scheduler ;; esac fi done fi
What the script does when it finds the SSD is to increase the block queue by a large amount, to set the deadline scheduler and to adjust the I/O-ratio. This improves the read speed and should help with the wear-leveling, too. -- If there is anything in your life you fear of losing, then keep your mouth shut once in a while. |
Xianthar
STK Scientific The Initiative.
|
Posted - 2009.05.31 16:07:00 -
[7]
technically there is no "official" support for ext4 in grub, support is there in grub2 but that is not declared stable. Theres also a google SoC project that has patches for grub 0.9x to support ext4 booting. Debian among other major distro's have merged the google SoC patch to support ext4, at least i think its the SoC patch they are using.
simplest option if your distro doesn't have this patched version of grub is to just leave /boot as ext3, its all grub cares about anyway.
|
Whitehound
|
Posted - 2009.05.31 16:58:00 -
[8]
Originally by: Xianthar technically there is no "official" support for ext4 in grub, support is there in grub2 but that is not declared stable. Theres also a google SoC project that has patches for grub 0.9x to support ext4 booting. Debian among other major distro's have merged the google SoC patch to support ext4, at least i think its the SoC patch they are using.
simplest option if your distro doesn't have this patched version of grub is to just leave /boot as ext3, its all grub cares about anyway.
I tried that but grub2 managed to cause a fsck-panic. Not exactly stable, no.
The use needs to be simple and straight forward, or else I risk fiddling with my Linux box when I do not have the time for it. I will wait, but ext4 itself looks good.
I am using a geometry for the SSD of 128 heads and 32 sector to get the 2nd partition aligned (just needs an even number for the start cylinder). I use 1k block for the 1st and 4k blocks for the 2nd partition. The difference in read speed this makes can actually be measured and it is more than 10%. I did not do any write tests.
Reiserfs was my favorite file system in the past, but you only need to read what Hans Reiser did and you will find yourself taking a big step away from it. Did you know that he bought books about crime scene investigations only a few days after his wife "disappeared"? The police found the books in his car. He even used the knowledge about the location of her remains to bargain the sentence from 1st to 2nd degree murder. I do not care how many people keep reiserfs alive, but I am sure that it will disappear.
There is word about btrfs now, which will have some of the features of reiserfs and a mode for SSDs. It is something to watch out for. -- If there is anything in your life you fear of losing, then keep your mouth shut once in a while. |
Solbright
Advanced Security And Asset Protection
|
Posted - 2009.06.01 01:59:00 -
[9]
Originally by: Whitehound I am using a geometry for the SSD of 128 heads and 32 sector to get the 2nd partition aligned (just needs an even number for the start cylinder).
Is that for real? Is the PC still really stuck with CHS parameters like that?
----- The Eve Client - A Love Story - The single biggest fix CCP ever did to Eve. Keep it up! |
Xianthar
STK Scientific The Initiative.
|
Posted - 2009.06.01 04:05:00 -
[10]
i don't see any reason the guy being an a$$ in real life takes anything away from the code he's written, its not like your paying for it and thus directly supporting the guy. other people have picked up maintenance of the file system, it will die out like other file systems in its generation, due to age.
just toss /boot on a separate partition make it like 100MB and format as ext3, then you can do whatever you want with the rest of the partitions.
|
|
Whitehound
|
Posted - 2009.06.01 07:51:00 -
[11]
Edited by: Whitehound on 01/06/2009 07:55:18
Originally by: Solbright
Originally by: Whitehound I am using a geometry for the SSD of 128 heads and 32 sector to get the 2nd partition aligned (just needs an even number for the start cylinder).
Is that for real? Is the PC still really stuck with CHS parameters like that?
No. The C/H/S parameters is a legacy from the very first hard disks and has turned into an abstraction for modern disks. It exists just for backwards compatibility.
SSDs do not have any physical parts, but they do have a block-sized access to their memory. The block size is 4k. When you create a partition on an SSD and it does not start at a 4K boundary, but is miss aligned then your OS may force the SSD to read two blocks when it requested only one.
The default C/H/S today is X/255/63, which likely will cause a partition to start anywhere but nicely aligned to 4K blocks. This is the only reason why I am using a setup with 128 heads and 32 sectors. I could have used anything, but it makes the creation of aligned partitions simple. The first partition on my SSD is miss aligned, but I also told the OS to use smaller blocks for accessing the disk (1K). The nicely aligned partition uses 4K blocks, which also matches the page size of the memory management btw. The speed difference this makes then is about 10%-15%.
So nothing is stuck. Only the nerd who tries to optimize everything is stuck with it. -- If there is anything in your life you fear of losing, then keep your mouth shut once in a while. |
Whitehound
|
Posted - 2009.06.01 08:00:00 -
[12]
Originally by: Xianthar i don't see any reason the guy being an a$$ in real life takes anything away from the code he's written, its not like your paying for it and thus directly supporting the guy.
He not an a$$, he is a murderer. We do have a word for people like him. Carrying on what he created is a form of support and I do not want to have a "MurderFS" on my computer, not in this sense. -- If there is anything in your life you fear of losing, then keep your mouth shut once in a while. |
Solbright
Advanced Security And Asset Protection
|
Posted - 2009.06.01 15:58:00 -
[13]
Whitehound: I think you've misunderstood me, the drive geometry has been irrelavant since the early 1980's when SCSI introduced block addressing. You appear to be using CHS terminology. Did block addressing never happen for the PC boot up process?
Is the PC still living in the 1970's?
----- The Eve Client - A Love Story - The single biggest fix CCP ever did to Eve. Keep it up! |
Whitehound
|
Posted - 2009.06.01 21:06:00 -
[14]
Originally by: Solbright ...
You must be one of those Mac user I have heard so many rumors about. I did not not misunderstand you, but you misunderstand me.
How much do you know about SSDs? I think that is where we should start discussing the use of C/H/S parameters, before I give in and tell you that you do need to worry about it when you do not have an SSD. -- If there is anything in your life you fear of losing, then keep your mouth shut once in a while. |
Ravow
Minmatar Cosmic Encounter
|
Posted - 2009.06.01 23:20:00 -
[15]
As mentionned, If you want to use non supported FS in grub, simply make a small /boot in something like ext2 and create your / with what you want! (With C/H/S trick to align partition except the first one, the small /boot only readed one time...)
Also align is not only for a faster reading, but it is for a faster write too... You know? writing one bit on unaligned partition = reading two block, erasing two block, writing two block...
And finaly, the only competitor for Reiser4 is btrfs, but it is still in devel and the format is still not frozen so no so recommanded. Reiser4 is still a very good file system. Yes Hans reiser killed is wife but the file system code and logic is very good and still very supported. ReiserFS is not MurderFS and we work hard to get it inside the kernel! If you call Reiser4 Murder4.. Then burn all IBM system you see on your way.. (Yes.. They sold a ton of punch-card system to help the ****s in there killing job and they producted some weapons too!)
|
Whitehound
|
Posted - 2009.06.02 00:16:00 -
[16]
Originally by: Ravow Then burn all IBM system you see on your way.. (Yes.. They sold a ton of punch-card system to help the ****s in there killing job and they producted some weapons too!)
No ..... thanks. -- If there is anything in your life you fear of losing, then keep your mouth shut once in a while. |
Solbright
Advanced Security And Asset Protection
|
Posted - 2009.06.02 00:43:00 -
[17]
Originally by: Ravow (With C/H/S trick to align partition except the first one, the small /boot only readed one time...)
I guess that confirms it. I assume this a legacy BIOS problem rather than a fundamental property of the spec ATA?
One more to the list of fixes the PC still needs.
----- The Eve Client - A Love Story - The single biggest fix CCP ever did to Eve. Keep it up! |
Whitehound
|
Posted - 2009.06.02 02:17:00 -
[18]
Originally by: Solbright I guess that confirms it. I assume this a legacy BIOS problem rather than a fundamental property of the spec ATA?
One more to the list of fixes the PC still needs.
Again, it exists only for backwards compatibility. It makes it simple to align a partition by that. It has nothing to do with how the drive is being addressed.
Did you understand that there is a small gain to be had with SSDs when a partition starts at a 4K boundary? Do you know what the size of a normal hard disk block is? -- If there is anything in your life you fear of losing, then keep your mouth shut once in a while. |
Solbright
Advanced Security And Asset Protection
|
Posted - 2009.06.02 12:26:00 -
[19]
Originally by: Whitehound Again, it exists only for backwards compatibility.
And here you are, in this very thread, being compatible with your backwards computers (plural).
Quote: It makes it simple to align a partition by that.
I would have thought that literally specifying, say, 2000 blocks for a 1MB partition was somewhat simpler than the bogus CHS figures I keep seeing presented.
Quote: It has nothing to do with how the drive is being addressed.
Really? Seems a lot like an address to me. Reminds me of segmentation. Boy, was that a load of drawn out bull****.
Evolution moves faster than the PC industry.
----- The Eve Client - A Love Story - The single biggest fix CCP ever did to Eve. Keep it up! |
Whitehound
|
Posted - 2009.06.02 12:31:00 -
[20]
Originally by: Solbright ...
Stop trolling. -- If there is anything in your life you fear of losing, then keep your mouth shut once in a while. |
|
Solbright
Advanced Security And Asset Protection
|
Posted - 2009.06.02 12:45:00 -
[21]
Hehe, is it trolling if it's true though?
----- The Eve Client - A Love Story - The single biggest fix CCP ever did to Eve. Keep it up! |
Ravow
Minmatar Brutor tribe
|
Posted - 2009.06.02 16:28:00 -
[22]
Edited by: Ravow on 02/06/2009 16:29:46 Yes principaly when it's not true.
Drive are drived with LBA. The C/H/S thing in fdisk is simply to fake a different geometry and this way we can align partition to 4k. When we pass fake CHS parameter, we can create partition normally and they will all be aligned except the first one (the small /boot). We user fdisk normally to create partition and we dont have to calcul the block nomber for each...
Finnaly, We use principally the first mesurement before the commercial modified it. 1K is not 2000 block but 2048.. so without a calculator, tell me how many for 112GO (4k aligned so you must put the good first block and the block count) :) Personnaly I prefer enter size of partition in MiB and GiB.
|
Solbright
Advanced Security And Asset Protection
|
Posted - 2009.06.02 23:36:00 -
[23]
Originally by: Ravow Drive are drived with LBA.
Then where is the block numbering? Why use CHS at all?
Quote: .. and this way we can align partition to 4k.
Not relevant for this discussion.
Quote: Finnaly, We use principally the first mesurement before the commercial modified it. 1K is not 2000 block but 2048..
Yep, sorry, was typing too quickly and didn't think. There is no such measurement as 1MB = 2000 blocks. 2000x512=1024000 bytes, which can only be named as 1000kB.
----- The Eve Client - A Love Story - The single biggest fix CCP ever did to Eve. Keep it up! |
Whitehound
|
Posted - 2009.06.03 01:52:00 -
[24]
Originally by: Solbright Then where is the block numbering? Why use CHS at all?
Quote: .. and this way we can align partition to 4k.
Not relevant for this discussion.
Why is it not relevant? You have not yet understood the problem with misaligned partitions and why we use CHS to get a partition aligned. -- If there is anything in your life you fear of losing, then keep your mouth shut once in a while. |
Xianthar
STK Scientific The Initiative.
|
Posted - 2009.06.03 03:43:00 -
[25]
Originally by: Whitehound Edited by: Whitehound on 01/06/2009 10:23:04
Originally by: Xianthar i don't see any reason the guy being an a$$ in real life takes anything away from the code he's written, its not like your paying for it and thus directly supporting the guy.
He is not an a$$, he is a murderer. We do have a word for people like him. Carrying on what he created is a form of support and I do not want to have a "MurderFS" on my computer, not in this sense.
if you truly believe this you should stop using linux, the DoD and NSA have made quite a few contributions to the kernel and they've killed a _few_ more people than Reiser did
|
Whitehound
|
Posted - 2009.06.03 08:37:00 -
[26]
Originally by: Xianthar if you truly believe this you should stop using linux, the DoD and NSA have made quite a few contributions to the kernel and they've killed a _few_ more people than Reiser did
You compare the DoD and NSA to murderers? I understand that you do not have any morals, because you cannot tell one thing from another. Just accept that I can tell the difference and that I do not want to use ReiserFS any more. Thank you. -- If there is anything in your life you fear of losing, then keep your mouth shut once in a while. |
Solbright
Advanced Security And Asset Protection
|
Posted - 2009.06.03 13:48:00 -
[27]
Originally by: Whitehound Why is it not relevant? You have not yet understood the problem with misaligned partitions and why we use CHS to get a partition aligned.
I know why you are doing it but I wasn't concerned with what you were trying to do. I was only pointing out how pathetic it is that CHS is still used to input that data.
I was actually wondering if the specific C, H and S values have any impact on the boot process beyond partition size. I.e.: Can the BIOS and/or bootloader and/or filesystem be confused or miscommunicate if those parameters are fiddled with while still not changing the size of the partition?
----- The Eve Client - A Love Story - The single biggest fix CCP ever did to Eve. Keep it up! |
Solbright
Advanced Security And Asset Protection
|
Posted - 2009.06.03 14:08:00 -
[28]
Originally by: Whitehound You compare the DoD and NSA to murderers?
Absolutely!
Obviously murder isn't DOD/NSA policy but individuals do take matters beyond policy all too often. Sometimes even with verbal consent in an attempt to contain an unverified and imagined threat. As they say, power corrupts. The result is blood on the hands of those in command and the institution they represent.
Transparency and accountability is the way forward.
I don't normally comment such topics so I'll leave it at that.
----- The Eve Client - A Love Story - The single biggest fix CCP ever did to Eve. Keep it up! |
Xianthar
STK Scientific The Initiative.
|
Posted - 2009.06.03 17:01:00 -
[29]
Originally by: Solbright
Originally by: Whitehound You compare the DoD and NSA to murderers?
Absolutely!
Obviously murder isn't DOD/NSA policy but individuals do take matters beyond policy all too often. Sometimes even with verbal consent in an attempt to contain an unverified and imagined threat. As they say, power corrupts. The result is blood on the hands of those in command and the institution they represent.
Transparency and accountability is the way forward.
I don't normally comment such topics so I'll leave it at that.
signed...and to a previous point, my morals are quite fine thank you, i just don't make special exceptions for those in a perceived position of power aka government agencies.
|
Whitehound
|
Posted - 2009.06.03 17:59:00 -
[30]
Edited by: Whitehound on 03/06/2009 17:59:38
Originally by: Solbright I.e.: Can the BIOS and/or bootloader and/or filesystem be confused or miscommunicate if those parameters are fiddled with while still not changing the size of the partition?
What sort of confusion are we talking about? The kind where you create two empty threads in a forum or the kind that we call a bug in the BIOS or the bootloader? -- If there is anything in your life you fear of losing, then keep your mouth shut once in a while. |
|
|
|
|
Pages: [1] 2 :: one page |
First page | Previous page | Next page | Last page |