I was trying to install PC-BSD but the installer freezes while copying files to the SSD. I tried to switch to other VTs to find out what was going on but the screen became corrupted.
The I tried installing DragonFlyBSD and the installer finished, probably only because there was less to install.
As I installed pkgs, disk errors started to appear about NCQ and trim.
I did some searching and found that many in-Windows-ok Samsung SSD have problems with filesystem corruption after ncq-enabled trim in linux. Some recent kernels have been patched to deal with this. I decided to test Lubuntu 15.04: it installed but after a few hours of intense-for-desktop disk activity, I discovered that dpkg-* binaries had been corrupted and were seg. faulting and apt-* scripts were failing. In syslog there were errors about NCQ.
So I reinstalled and immediately updated the from 3.19.0-15-generic (built sometime in April 2015, definitely before the Samsung NCQ problem was found) to the latest 3.19.0-28-generic before repeating the disk workout. (The testing is apt-get installing texlive while copying 128GB from my external drive to the SSD at about 1GB/min.)
Everything seemed fine but there were lots of messages in syslog:
Adding the boot parameter libata.force=noncq stopped the warnings about NCQ and everything seems fine with ncq disabled:
Is there a way to disable native command queue at boot in FreeBSD? Are there other problems with this drive/controller and BSD? It could be that the drive is just a dud?
Oh, and what the hell is "horkage?"
The I tried installing DragonFlyBSD and the installer finished, probably only because there was less to install.
As I installed pkgs, disk errors started to appear about NCQ and trim.
I did some searching and found that many in-Windows-ok Samsung SSD have problems with filesystem corruption after ncq-enabled trim in linux. Some recent kernels have been patched to deal with this. I decided to test Lubuntu 15.04: it installed but after a few hours of intense-for-desktop disk activity, I discovered that dpkg-* binaries had been corrupted and were seg. faulting and apt-* scripts were failing. In syslog there were errors about NCQ.
So I reinstalled and immediately updated the from 3.19.0-15-generic (built sometime in April 2015, definitely before the Samsung NCQ problem was found) to the latest 3.19.0-28-generic before repeating the disk workout. (The testing is apt-get installing texlive while copying 128GB from my external drive to the SSD at about 1GB/min.)
Everything seemed fine but there were lots of messages in syslog:
Code:
ata1.00: failed to get NCQ Send/Recv Log Emask 0x1Adding the boot parameter libata.force=noncq stopped the warnings about NCQ and everything seems fine with ncq disabled:
Code:
cat /proc/cmdline
BOOT_IMAGE=/boot/vmlinuz-3.19.0-28-generic
root=UUID=8e3b1891-8756-4db4-b77d-ed966b7ae738 ro
libata.force=noncq quiet splash vt.handoff=7
dmesg | grep noncq
ata1.00: FORCE: horkage modified (noncq)
lspci | grep SATA
00:1f.2 SATA controller: Intel Corporation
NM10/ICH7 Family SATA Controller [AHCI mode] (rev 02)Is there a way to disable native command queue at boot in FreeBSD? Are there other problems with this drive/controller and BSD? It could be that the drive is just a dud?
Oh, and what the hell is "horkage?"