Date: Wed, 19 Oct 2022 15:31:24 +0200 From: Borja Marcos <borjam@sarenet.es> To: "freebsd-scsi@freebsd.org" <freebsd-scsi@FreeBSD.org> Subject: Broadcom AERO SAS adapter: JBOD personality bug Message-ID: <E4C5B6D3-550A-4B43-B7A0-59C72FA9149E@sarenet.es>
next in thread | raw e-mail | index | archive | help
Hi, Hope someone from Broadcom is still lurking here ;) I have seen a problem with a new Broadcom Aero SAS adapter on FreeBSD. I = first detected it on TruNAS 12 but the issue affects FreeBSD 13.1 as well. The card came with RAID firmware installed (dealer fault, but with crazy = delivery times now I am afraid we must swallow it) but fortunately it has =E2=80=9CPersonalities=E2=80=9D.=20 So, if I set it up in =E2=80=9CJBOD personality=E2=80=9D it does behave = like a plain HBA. Good! However, I have noticed a problem in the JBOD personality. I am not sure = whether it is a firmware bug, a design error or a problem with the mrsas driver although after a quick look at the = source code I don=E2=80=99t think the driver can solve this. After a disk failure (for example, pulling a disk from the backplane) = the controller will mark it as UNCONFIGURED BAD. Which is a dead end that may require a system reboot = or at least using a tool such as StorCli (if it is possible, I haven=E2=80=99t verified that yet). If this is a design decision I THINK THIS IS A MISTAKE. If the JBOD = personality is equivalent to an HBA (it should be!) the controller firmware should not interfere with decisions made by = the operating system itself. As for the mrsas driver, I am not sure, is there a possibility to add a = tunable so that can ignore the =E2=80=9Cunconfigured badness=E2=80=9D and proceed?=20 The controller is a Broadcom AERO 10E2 Firmware: 5.180.02-2512 NVDATA version: 5.1800.21-0642 Thanks! Borja Marcos.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?E4C5B6D3-550A-4B43-B7A0-59C72FA9149E>