Date: Sun, 5 Mar 2017 22:01:32 +0100 From: Dirk-Willem van Gulik <dirkx@webweaving.org> To: Warner Losh <imp@bsdimp.com> Subject: Re: PCI speed Message-ID: <18D26C89-0202-40F6-9396-586ADA79C67C@webweaving.org> Resent-Message-ID: <201703052111.v25LBunt001786@weser.webweaving.org> In-Reply-To: <CANCZdfqMmkJyyHBMB7zS72gBjYcR_7kGxrxHmtq83LbL-SmFdA@mail.gmail.com> References: <03745A05-DE0E-4071-B432-BFE58C3A7E16@webweaving.org> <CANCZdfqMmkJyyHBMB7zS72gBjYcR_7kGxrxHmtq83LbL-SmFdA@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
> On 3 Mar 2017, at 23:52, Warner Losh <imp@bsdimp.com> wrote: >=20 > On Fri, Mar 3, 2017 at 3:34 PM, Dirk-Willem van Gulik > <dirkx@webweaving.org> wrote: >> Forgive me my ignorance - but for some cards - pciconf(8) nicely = lists the speed of the bus: >>=20 >>> ciss0@pci0:12:0:0: class=3D0x010400 card=3D0x3243103c = chip=3D0x323a103c rev=3D0x01 hdr=3D0x00 >> ... >>> cap 10[70] =3D PCI-Express 2 endpoint max data 128(256) RO NS >>> link x4(x8) speed 5.0(5.0) >>=20 >> and that matches exactly with what it is. While for other cards it = does not seem to report that: >=20 > So you have a x8 card in a x4 slot. Hope that's OK. It's also > reporting errors on the PCIe link level. That doesn't sound too > correct. It ended up being quite a rat-hole - further examination showed that = some on motherboard soldered P410 and NC382i ethernets where reporting = just as oddly! >>> mpt4@pci0:7:8:1: class=3D0x010000 card=3D0x10b01000 = chip=3D0x00301000 rev=3D0x08 hdr=3D0x00 >> .... >>> cap 07[68] =3D PCI-X 64-bit supports 133MHz, 2048 burst read, 8 = split transactions >>=20 >> and in fact seems to mis report the bus - this is a PCIe Gen 2 x4 = bus with a x8 Connector Width >> holding a LSI 22320SE Ultra320 SCSI dual channel PCIe X4 card. >=20 >> How should one interpret this ? >=20 > That's odd. I'd have expected it to report correctly... It does for us = at work. > What version of FreeBSD? FreeBSD 11.0 - p7. The HW are fairly plain DL385 G7 and G8 frames with = AMD cpu's and two risers (One bus x8, connector 16 and two x4bus and x8 = connector). Two hold an P410i/2048MB cards; two P222/512Mb cards and one = LSILogic 1030 Ultra4 Adapter. We've since looked at multiple DL385's and found the results to be a = total mess; with reports even differing between the first boot after a = hard power cycle and the next boot(s). Googling & Escalation within HPE = suggests that it is one large firmware mess with conflicting reports = (P2222 cards clashing with the Qlogic needs).=20 So I guess this is a rathole best left closed. Old HP kit is essentially = not worth it anymore given their serial#/paid-support burden. Dw.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?18D26C89-0202-40F6-9396-586ADA79C67C>