Hello, please cc me I'm not subscribed. Beginning with commit 64d513ac "scsi: use host wide tags by default" my system will ONLY boot with CONFIG_SCSI_MQ_DEFAULT=y. Tested through git-stable master (0adb328 v4.16). It never gets far enough to log anything, but here are a few boot messages that I copied by hand (so do not rely on their accuracy): ________________________________________________ Exception Emask 0x0 SAct 0x1 SErr 0x0 action 0x6 Frozen ... ata2: SRST failed (errno = -16) ata2: reset failed, giving up ata2: nv: skipping hard reset on occupied port ... ata2.00: disabled ata2: EH complete ________________________________________________ It doesn't always happen in the same place, but 95% of the time it freezes late into run level 3 when the GTK immodule cache is updated. Affected hardware messages (from a successful boot using SCSI_MQ_DEFAULT=y): [ 0.184013] SCSI subsystem initialized [ 0.184207] libata version 3.00 loaded. [ 0.434160] Block layer SCSI generic (bsg) driver version 0.4 loaded (major 250) [ 1.357484] sata_nv 0000:00:0e.0: version 3.5 [ 1.357885] sata_nv 0000:00:0e.0: Using SWNCQ mode [ 1.358785] scsi host1: sata_nv [ 1.359070] ata2: SATA max UDMA/133 cmd 0xe80 ctl 0xe00 bmdma 0xe008 irq 21 [ 4.848023] ata2: SATA link up 3.0 Gbps (SStatus 123 SControl 300) [ 4.867282] ata2.00: ATA-7: SAMSUNG HD502HI, 1AG01118, max UDMA7 [ 4.867723] ata2.00: 976773168 sectors, multi 16: LBA48 NCQ (depth 31/32) [ 4.887272] ata2.00: configured for UDMA/133 [ 4.888439] scsi 1:0:0:0: Direct-Access ATA SAMSUNG HD502HI 1118 PQ: 0 ANSI: 5 [ 4.890456] sd 1:0:0:0: [sdf] 976773168 512-byte logical blocks: (500 GB/466 GiB) [ 4.890926] sd 1:0:0:0: [sdf] Write Protect is off [ 4.891370] sd 1:0:0:0: [sdf] Mode Sense: 00 3a 00 00 [ 4.893682] sd 1:0:0:0: [sdf] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA [ 4.896117] sdf: sdf1 sdf2 Thanks, William
