Date: Thu, 17 Sep 2015 10:30:12 -0600 From: Warner Losh <imp@bsdimp.com> To: Bruce Simpson <bms@fastmail.net> Cc: Warner Losh <imp@freebsd.org>, src-committers <src-committers@freebsd.org>, svn-src-projects@freebsd.org Subject: Re: svn commit: r287876 - in projects/iosched/sys/cam: . ata scsi Message-ID: <CANCZdfo8K3R-QiTrTqyjS5YQZ1rt8KTZWPYi4p-tbxAEDWX-9A@mail.gmail.com> In-Reply-To: <55FAD739.1050800@fastmail.net> References: <201509162215.t8GMFp1b023705@repo.freebsd.org> <55FAB5DD.20609@fastmail.net> <CANCZdfoqAFmJEpnpzWx%2BtFQG5GN9rDTTnpfr=vpVUVGzDMakYg@mail.gmail.com> <55FAD739.1050800@fastmail.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, Sep 17, 2015 at 9:07 AM, Bruce Simpson <bms@fastmail.net> wrote: > On 17/09/15 15:16, Warner Losh wrote: > >> This isn't even in -current yet, but this will work in 10.x. It's >> dependent, at the moment, >> on the I/O scheduler work I've done in this project branch. There's >> nothing intrinsic about >> that. >> > > That is good news; the bug specifically affected NCQ'd TRIM commands. > Performing TRIM manually is an admin headache and gets old, fast. We don't do NCQ Trims today. When I deployed, I had massive corruption on M500's that I developed for after a few hours to days. We do have non-NCQ Trims and those work just fine on these drives. However, they serialize other reads and writes. I hope to forward port this to -current soon. Warner
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CANCZdfo8K3R-QiTrTqyjS5YQZ1rt8KTZWPYi4p-tbxAEDWX-9A>