Almost exactly a year ago (give or take an hour), we disabled AEN handling in
mfii(4)
because it didn't work on SAS2208 controllers. I finally got around to looking
into this
and found we were just missing one magical offset in one of the command
structures. The
diff below add this and re-enables AEN processing.
I've tested on the following:
mfii0 at pci3 dev 0 function 0 "Symbios Logic MegaRAID SAS2208" rev 0x01: msi
mfii0: "PERC H710P Mini", firmware 21.0.2-0001, 1024MB cache
mfii0 at pci1 dev 0 function 0 "Symbios Logic MegaRAID SAS3108" rev 0x02: msi
mfii0: "PERC H730 Mini", firmware 25.4.0.0015, 1024MB cache
Could someone test on a SAS3.5 controller (megaraid 34xx or 35xx)? I don't
have access to
one currently. All you need to do is boot the resulting kernel and check mfii0
isn't
generating interrupts constantly. Tests on other hardware would also be useful.
Index: mfii.c
===================================================================
RCS file: /cvs/src/sys/dev/pci/mfii.c,v
retrieving revision 1.46
diff -u -p -r1.46 mfii.c
--- mfii.c 2 Feb 2018 11:24:37 -0000 1.46
+++ mfii.c 11 Feb 2018 02:34:50 -0000
@@ -630,22 +630,18 @@ mfii_attach(struct device *parent, struc
mfii_syspd(sc);
-#ifdef notyet
if (mfii_aen_register(sc) != 0) {
/* error printed by mfii_aen_register */
goto intr_disestablish;
}
-#endif
/* enable interrupts */
mfii_write(sc, MFI_OSTS, 0xffffffff);
mfii_write(sc, MFI_OMSK, ~MFII_OSTS_INTR_VALID);
return;
-#ifdef notyet
intr_disestablish:
pci_intr_disestablish(sc->sc_pc, sc->sc_ih);
-#endif
free_sgl:
mfii_dmamem_free(sc, sc->sc_sgl);
free_requests:
@@ -851,6 +847,7 @@ mfii_dcmd_start(struct mfii_softc *sc, s
io->function = MFII_FUNCTION_PASSTHRU_IO;
io->sgl_offset0 = (uint32_t *)sge - (uint32_t *)io;
+ io->chain_offset = io->sgl_offset0 / 4;
htolem64(&sge->sg_addr, ccb->ccb_sense_dva);
htolem32(&sge->sg_len, sizeof(*ccb->ccb_sense));