[Bug 250924] Bhyve AHCI disk controller regression due to r364334

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=250924

            Bug ID: 250924
           Summary: Bhyve AHCI disk controller regression due to r364334
           Product: Base System
           Version: 12.2-RELEASE
          Hardware: amd64
                OS: Any
            Status: New
          Severity: Affects Some People
          Priority: ---
         Component: bhyve
          Assignee: virtualization at FreeBSD.org
          Reporter: freebsd at gulag.ch

Created attachment 219426
  --> https://bugs.freebsd.org/bugzilla/attachment.cgi?id=219426&action=edit
diff patch against r364334

Since https://svnweb.freebsd.org/base?view=revision&revision=364334 Solaris and
Illumos guests are no longer able to detect the Bhyve AHCI disk controller and
the following error is reported:

---
SATA disk device at port X does not support UDMA
---

https://github.com/illumos/illumos-gate/blob/master/usr/src/uts/common/io/sata/impl/sata.c#L12752

The culprit seems to be the missing ATA flag 88 at offset 53 in the parameter
buffer and this flag is indeed no longer set in 12.2-p0:

RELEASE-12.1:

buf[53] = (1 << 1 | 1 << 2);

RELEASE-12.2:

ata_ident->atavalid = (ATA_FLAG_54_58 | ATA_FLAG_64_70)

sys/sys/ata.h:
#define ATA_FLAG_54_58                  0x0001  /* words 54-58 valid */
#define ATA_FLAG_64_70                  0x0002  /* words 64-70 valid */
#define ATA_FLAG_88                     0x0004  /* word 88 valid */

---

Setting the old parameters solves the problem (at least for Solaris/Illumos):

ata_ident->atavalid = (ATA_FLAG_64_70 | ATA_FLAG_88)

The attached patch reverts these parameters and thus restores the old behaviour
for both controllers (disk and cdrom).

The question is whether this change is intentional or simply a bug.

-- 
You are receiving this mail because:
You are the assignee for the bug.