Debian Hardware :: NCQ (queued Trim) With Samsung 850 Pro SSD
Aug 29, 2015
I just bought a Samsung 850 Pro 250GB SSD drive and have since then found out that Linux has/had a faulty implementation of queued trim (which is a synonym for NCQ I think). I'm using kernel 4.1.0 (the default selected by the Debian 8.1 installer) and I've read that Samsung drives are blacklisted for queued trim by the recent kernels. However, the output of dmesg contains the following lines, with and without TRIM enabled for the drive (i.e. with or without the "discard" option in fstab for the drive):
Code: Select all[ 1.012287] ata6.00: 1953525168 sectors, multi 16: LBA48 NCQ (depth 31/32), AA
[ 1.013438] ata5.00: failed to get NCQ Send/Recv Log Emask 0x1
[ 1.013440] ata5.00: 488397168 sectors, multi 1: LBA48 NCQ (depth 31/32), AA
[ 1.015037] ata5.00: failed to get NCQ Send/Recv Log Emask 0x1
Since the depth is not 1 but rather 31, it looks like NCQ is enabled. I've disabled TRIM on the drive for now, and whether it's safe to enable TRIM or not.
does Debian Lenny 5.0.4 stable supports SSD TRIM? If not, what package i need to add if i buy a new SSD and dont want to loose Hard Disk Performance over time?
I'm looking for a way to launch "TRIM" commands to a SSD drive myself, from a c++ code I'm going to write, for flash erasing an SSD drive (and not to wait for others things that "should do it automatically in some circumstances if this or this, this and this have been enabled and [..]" but will never tell me if it worked or not
I know there is thousands of complicated ways to test and check if it worked, and also software that needs money to do so, that's why I just want to call myself the TRIM functions and read the return value (like true or false) in order to know if it worked !
Where I can find the c++ call that could permit me to do so ? I heard about the GLibC that gives a way to access every user space function related to Linux Kernel (poll, select and others) as standard c++ functions, I suppose that, if there is a way, it will be on the GLibC but how to find it ?
I've a mailserver based on postfix. I'm queueing all the emails that have to be sent. After that I'm managing the queue with the output of mailq.
But in this way I'm just able to have this details of the queued mails: sender, receiver, size, time. There is a way to see also the subject of that emails?
I have a netbook with a 150GB hard drive in it. A while ago I dropped my netbook and it wouldn't boot. Had a "failed command: READ FPDMA QUEUED" error at startup. So naturally I went and reinstalled. Reinstalling worked fine and I had my computer back. What I didn't account for was the fact that anytime it lost power and didn't shut down properly, I'd get the same error. I can't seem to recover my data either. Can't get any live disks to mount the hard drive after this error but for some reason I can still reinstall Crunchbang (essentially debian). Now I know there are bad sectors or something because when I try to use the disk that came with the netbook with an image of Windows XP that writes to the entire hard drive it throws an error but linux will avoid this I suppose when installing.
Now all of this is to say, being a laptop, several times I have left it on overnight when I fall asleep working or have to run into to town and it'll die and then I lose all my data, rather irritating. What's causing this error? What can I do to prevent it besides getting a new hard drive? If I can't prevent it from happening in the future is there a way to recover the data off the hard drive? Here's the section of the boot log that contains the error: [url]
I like ubuntu alright, but prefer Debian overall. Before I reinstall Debian I was wondering if during the install if the latest release would automatically recognize the Samsung ML 1665 as the latest ubuntu release did. If not, how does one get the printer installed.
I had an ubuntu 11.10 installation that was no longer supported and was low on disk space, so I bought a new disk and installed debian wheezy on it and made it the boot disk.
The machine had/has a samsung e2020 monitor, and ubuntu and the new debian install both recognized it as 1600x900 resolution. I ran the proprietary nvidia driver with ubuntu but did nothing at all with the debian ... it just worked.
The power supply on the machine conked out and I took it ro be replaced and dug out an old, old machine to use while the newer one was gone. Debian would not install on it. But ubuntu 11.10 did, but displayed at 1024x768 on the samsung which it recognized as a 'VGA'. It was slow, but it worked. When the newer machine came back I reinstalled it ... but its debian 7.6, which used to know the samsung at 1600x900 resolution, now runs it at 1024x768, just as the old box had!
It must be some change made to some firmware in the samsung itself? It's not very pleasant.
I've tried to install the nvidia driver , but no luck there. I see now that debian uses something called 'nouveau' to run the proprietary nvidia chip, and I'm all for that ... but it used to be able to run it at 1600x900 and now it's forgotten how.
Here they are ...
Code: Select alljfl@wsdeb:~$ xrandr Screen 0: minimum 320 x 200, current 1024 x 768, maximum 4096 x 4096 VGA-1 connected 1024x768+0+0 (normal left inverted right x axis y axis) 0mm x 0mm 1024x768 60.0* 800x600 60.3 56.2 848x480 60.0 640x480 59.9
[Code] ....
So I've gotten lucky, haven't I? I don't know if the display will still be ok when I restart ... so I'll save a copy of the above and try to script it, if I must, on reboot/relogin.
Environment: Debian Lenny | Samba/CUPS | Network: 25 Workstations (WINXP SP3) | Printer: Samsung ML2010 (on server) That�s it. Problem: All workstations can print, except two. I formated the two workstations, installed everything again and nothing!
As we know the linux driver supplied by Samsung for the ml-1740 printer did not work for Debian. Debian has removed the ml-1710 and ml-1750 drivers from its database of Samsung drivers, in my case, the driver for the ml-1710 worked well but when I try to download it now from the internet, I cannot get it to work, somewhere in the install and after some work has been done, I get a message "package not found" although the full path of the package is copy(ed) from the directory where it is saved and past(ed) after "dpkg install ".
I recently bought a new Samsung netbook N310 which has windows xp home edition on it. My CPU is like this: Intel Atom CPU N270 1.6GHz. I now want to install Debian Lenny 5.06 with GUI, while keep the windows xp partition for later use.
how I can boot directly from a 2GB USB drive, instead of having to modify grub etc. Is there something in Debian that is equivalent to the Universal USB Installer in Ubuntu?
Trying to set to native resolution of 1360x768 with error "Failed to get size of gamma for output default":
Code: Select allerik@DRAGON:/var/log$ xrandr xrandr: Failed to get size of gamma for output default Screen 0: minimum 640 x 480, current 1024 x 768, maximum 1360 x 768 default connected 1024x768+0+0 0mm x 0mm 1024x768 61.00* 800x600 61.00
[Code] ....
Found these lines in /var/log/kern.log stating "unknown Kepler chipset":
Code: Select allJul 26 17:35:34 DRAGON kernel: [ 15.510099] nouveau ![ DEVICE][0000:03:00.0] unknown Kepler chipset Jul 26 17:35:34 DRAGON kernel: [ 15.510148] nouveau E[ DEVICE][0000:03:00.0] unknown chipset, 0xb06070b1 Jul 26 17:35:34 DRAGON kernel: [ 15.510191] nouveau E[ DRM] failed to create 0x80000080, -22 Jul 26 17:35:34 DRAGON kernel: [ 15.510419] nouveau: probe of 0000:03:00.0 failed with error -22
Google search yielded this bug, which affected GT 730 graphics cards, apparently patched in Linux 3.19:
[URL] ....
Assuming this is the problem, does this mean I have to upgrade to Testing or install proprietary drivers?
I often use popcorn time [URL] .... to watch some movies. It allows me automatically to stream a movie to my samsung smart TV (which is connected to the network, but I do not have anything like chromecast connected to it).
Now, seen that it is possible for popcorn time to stream to my TV, how can I do it myself e.g. in the case of a video file or some pictures which are on my PC?
I just installed a minimal Debian Testing on a USB stick to be used with my Samsung N220 netbook. I installed X and noticed that about half of the keys are not working. I fail to find any configuration tips on google
A relative of mine who uses Debian 6.0.2 has bought a Samsung HMX-H300 Camcorder which records videos in:
1. H.264 (MPEG4 part10/AVC) format
2. High Definition Video (HD-VIDEO)
3. Standard Definition Video.
I notice that the PC requirement for the camcorder's bundled software, apart from the usual Windows XP SP2/Vista/Windows 7, lists an Intel Core Duo 1.66 minimum CPU and an NVIDIA GeForce 8500 or higher or ATI HD 2600 series or higher.
My relative has an old Intel(R) Celeron(R) CPU 2.30GHz computer which only has an AGP GeForce FX 5200 but after copying the videos from the Memory Card to /home it is just about possible to play them using Totem Movie Player and VLC but, although the audio is OK the video is painfully slow or almost stationary.
Is there a way to view the videos recorded in these formats with this hardware? Can they be converted to another format? Or should he consider buying a better computer?
Just wanted to check, would the kernel in Jessie blacklist queued trim for the Samsung 840 EVO for compatibility with firmware EXT0DB6Q as per similar situations that have occurred with other drives?
Have 4xSamsung F4 HDDs in an Ex503, Raid 05 - don't know how to format them to get maximum speed out of them. I've read [URL] tor_issues and other things but guess i am too dumb. Some claim using GPT Partition table and gparted and i shouldn't have any problems because it's automatically doing what i want, but i am not sure about that.
I was running it portable on a 4gb usb drive via virtual box which worked great unless I used a computer that had virtualbox on it. The portable version would remove files that the original version needed. I decided to try Qemu. The problem is that the portable version only allows 1gb of space to install regardless the size of a usb drive. so the question is: Is there was a way to trim ubuntu 9.10 down to less than 1gb of disk space? If not is there another portable emulator that could be used instead? the only uses for this are for the use of evolution and opera in a linux environment.
We're planning to install CentOS on a new server that will be based on two Solid-State Disks in a RAID1 configuration. The RAID management will be handled by a hardware Adaptec controller. I've learnt that the TRIM function for SSD drives is supported starting from kernel 2.6.28.
Ok, Im likely to buy SSD 128gb drive and install Fedora 14 on it...but I have a question. If I go with ext4 will I get TRIM to work out of the box? Whats the status of kernel support on trim?
I have an OCZ Vertex SSD...the 30GB variety. recently I upgraded my kernel to 2.6.33. I have read that that is all that is needed in order for trim to just start working. Is there a way that I can verify that TRIM is working?I'm debating getting the Vertex2 100GB model because I love my SSD now, but would love to have a little more storage space. I just don't want to buy it without TRIM working.
I followed this tutorial [URL] to enable my TRIM support and now my fstab looks like this:
Quote:
# /etc/fstab: static file system information. # # Use 'blkid -o value -s UUID' to print the universally unique identifier # for a device; this may be used with UUID= as a more robust way to name # devices that works even if disks are added and removed. See fstab(5).
[Code].....
but when I follow the second half of the tutorial http://lightrush.ndoytchev.com/rando...bledandworking to check and make sure it's working I don't get all zeros like I'm supposed. I tried waiting a while like I saw in another tutorial as well and check it again.. Still random numbers and no zeros.
First system specs. Gateway Solo Pro 9300, 433Mhz, 288megs. OS Slackware 13.1 32-bit and KDE.
As I get more comfortable with Linux I find that I am doing more waiting for the computer to catch up.
Would it be better to trim down KDE or change to a lighter GUI? If I change to a lighter GUI will I have to reinstall things I have installed under KDE?
Does CentOS 5 support the ATA TRIM command for use with SSDs? Is the support automatic by default, or would I need to do something specific in order to enable the TRIM feature be used?
I am trying to set up an atom D525 low power PC 64-bit with a 40 GB solid state disk drive. Is it possible to specify ext4 during the install for proper SSD suupport? I read somewhere that after install I can place a -discard line in fstab to enable trim.
Edit: should I have asked this in the x86_64 forum, as I was planning on installing 64 bit?
Does Suse have any software based TRIM support available or that is in the works?I know there is to a degree new firmware becoming available that supports TRIM (although some are pretty rudimentary) but windows 7 does a fairly good job software side and upgrading the firmware occasionally involves the unbridled joy that is wiping the drive and starting again.
There are not SSDs with TRIM support available in my region that fit into my laptop (1.8", IDE, ZIF). I'm running Ubuntu 10.10.
Most articles (or questions on superuser) I've come across concering TRIM (or the lack thereof) date back to 2009, when not many SSDs with TRIM support were available and OS support was still very fresh.
I'm interested in the current situation, but I couldn't find too much information about it.
What are currently the "best practices" for using an SSD without TRIM under Linux? I've read about the wiper script included with hdparm. Do I understand correctly that I could use this to free unused blocks, e.g. by running it once a month? Some sources state that HFS+ (the default-filesystem of Mac OS X) doesn't suffer as badly from lack of TRIM as other filesystems. How about linux filesystems? Are there filesystems that are better suited for SSDs without TRIM than others?
My partition /dev/sda3 on an SSD drive doesn't contain any filesystem, but it contains garbage. How do I do a TRIM/DISCARD operation on the whole partition?