Adaptec Series 6 Raid Cards not working because of kernel bug

[stormi]

See discussion at https://xcp-ng.org/forum/topic/3979/kernel-bug-causes-adaptec-series-6-raid-cards-to-not-work

There's a provided patch, though linked mailing list message also mentions a follow-up patch.

XCP-ng releases : probably 8.0 to 8.2+

[stormi]

Current status: the provided patches never got merged upstream and have a larger scope than just the bug above. kernel-alt also has the bug so it still exists in upstream kernel. One would need to find/contribute a better patch.

[0nelight]

Kernel-Mailing List and Thread responsible for repairing the faulity 2017 Commit 395e5df79: https://lore.kernel.org/linux-scsi/20190819163546.915-2-khorenko@virtuozzo.com/

[0nelight]

Current Status in 8.2.1:

Package	XCP-ng 8.2	XCP-ng 8.2 alternate driver	XS 8.2	XS 8.2 CU1	XS driver disk
microsemi-aacraid	(built-in) 1.2.1[50877]-custom		(built-in) 1.2.1[50877]-custom	idem	1.2.1.60001

xcp-ng-8.2.1-20231130.iso
SHA256: 108aa7144d1f5dd169a1b99ca70b510527136c549f2e3c3c1707b619e17ee1b6
RAID bus controller [0104]: Adaptec Series 6 - 6G SAS/PCIe 2 [9005:028b] (rev 01)
-----
dmesg:
[ xx:xxxxx] aacraid: Host adapter abort request.
[ xx:xxxxx] aacraid: Outstanding commands on (X,X,XXX,X):
...

Current Status in 8.3.0

Package	XCP-ng 8.3	XCP-ng 8.3 alternate driver	XS 8 base	XS 8 earlyaccess	XS driver disk
microsemi-aacraid	1.2.1.60001-1		1.2.1.60001-1		1.2.1.60001

xcp-ng-8.3.0.iso
SHA256: 49b6143d1bbb1fd0bb1f7ce873817190c1f2ca394c258ce7b1f8e13a75c72674
RAID bus controller [0104]: Adaptec Series 6 - 6G SAS/PCIe 2 [9005:028b] (rev 01)
-----
dmesg:
[xx:xxxxx] aacraid 0000:03:00.0: AAC: aac_eh_abort: Host adapter abort request (X,X,X)
[xx:xxxxx] aacraid 0000:03:00.0: AAC: aac_eh_abort: Timed out Command: 12 00 00 24 00 00 00 00 00 00 00 00 00 00 00
[xx:xxxxx] aacraid 0000:03:00.0: AAC: aac_fib_debug_print: FIB(0) = 000000005c443e4 : be400020 Command = 601 XferState = 000832ad Wait Time = 20 Sec

Notes

Usage as a non-boot Storage-Device: Install XCP-NG without the Raid-Card being inserted into the mainboard and then replace the *.ko file (compile it yourself). After that you can Insert the Raid-Card.

Usage as a XCP-Boot Device: Modify the ISO-Image because the Raid-Card leads to very long hangs during the Installer-Boot process so you can't even do anything in the command line.

I successfully created an custom-driver-iso which I can load via the installer. My Problem is that I don't know how I can de-blacklist the kernelmodule aacraid after loading the correct aacraid via the installer so I can continue with the Install. Should I patch the Device-ID out of https://github.com/xcp-ng-rpms/microsemi-aacraid linit.c:262 so it doesn't load the faulty driver for this card in 8.3? At least then, the command line will still be usable and the installer would not hang.

[0nelight]

I am very sure that the following would fix Series-6 Cards in XCP 8.3. This would restore the logic for Series-6 as in the Latest Supported Driver from the Adaptec Support Page (Driver 52011), without changing anything for the other Cards. Will update here as soon I have tested this :-)

diff --git a/comminit.c b/comminit.c
index 0c44ba8..d035e18 100644
--- a/comminit.c
+++ b/comminit.c
@@ -727,7 +727,7 @@ int aac_init_adapter(struct aac_dev *dev)
                }
        }

-       if (aac_is_srcv(dev))
+       if (aac_is_src(dev))
                aac_define_int_mode(dev);

 #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,11,0))

cat aacraid.h

static inline int aac_is_src(struct aac_dev *dev)
{
    u16 device = dev->pdev->device;
    if (device == PMC_DEVICE_S6 ||
        device == PMC_DEVICE_S7 ||
        device == PMC_DEVICE_S8)
        return 1;
    return 0;
}

static inline int aac_is_srcv(struct aac_dev *dev)
{
    u16 device = dev->pdev->device;
    if (device == PMC_DEVICE_S7 ||
        device == PMC_DEVICE_S8)
        return 1;
    return 0;
}

comminit.c from Driver 52011

        }
    }

    if (dev->pdev->device == PMC_DEVICE_S6 ||
        dev->pdev->device == PMC_DEVICE_S7 ||
        dev->pdev->device == PMC_DEVICE_S8 ||
        dev->pdev->device == PMC_DEVICE_S9)
        aac_define_int_mode(dev);

[stormi]

Thanks, keep us informed of the results!

[0nelight]

It works 😃 Installer doesn't hang anymore, no Error-Messages and Raid-Volume is visible in the Installer 😍

I managed to create a patch RPM and modify the ISO - thanks to the well written guides :-) I created a Pull-Request. I am very beginner in this! I am not sure if I should edit the Specfile further. For example, I think it would be better if the Version number one sees when doing modinfo aacraid should indicate that this is the Series6 fix.