Huge hpet0 interrupt storm, but only if HGST HTS721010A9E630 HDD is present at the boot _and_ formatted with recognized fs (??)

[Jakub Lach]

Hello, I would like to share symptoms I've encountered
to hopefully spare somebody the headache I have now. 

HGST HTS721010A9E630 (HGST Travelstar 7K1000) is good 
value now, it's a 1TB 2,5" 7200 rpm HDD with 100 MB/s+ 
reads and writes. I've hoped to use it as a SSD backup.

Environment:

FreeBSD 11.0-STABLE #0 r309672 amd64
Lenovo Thinkpad T400, latest official BIOS
HGST HTS721010A9E630 is the problematic HDD, JB0OA3J0 firmware

Description:

Laptop without added HGST disk works fine. Situation does not 
change after adding the disk- it is recognized correctly, diskinfo
reports correct throughput, after enabling SMART, smartctl 
reports are all clear. You can now reboot the system with the 
disk attached without any problems.

However, after the disk is gpt formatted to recognizable FS (UFS2+SU, 
no journaling, 1  partition freebsd-ufs  (932G), I have used @wblock guide), 
as soon as you reboot the machine, there is a huge interrupt storm 
with hpet0 (vmstat -i) on the restart, which stalls the computer. It does
not 
matter if said disk is mounted on reboot or not. All it needs is a mere 
presence of a disk with recognizable FS. 

Nonetheless, if such disk were to be added mere seconds after a booting, 
it would work fine. As verified by vmstat -i. If I were to manually
eject/reinsert 
the disk after each reboot, I could use it without problems.

* I have photos of the interrupt storm, though it's just irq20 hpet0 rate
120-570
normally and about  305000 rate with the storm. The storm only happens if
the disk is already attached during the reboot process. The irq attached to
disks
are the same between storm/no storm scenario. All that changes is hpet0 
interrupt rate.

* I can use two disk at the same time because I'm using the the bay in place 
of optical drive, however I've swapped some disks around and I'm positive 
the problem lies with HGST disk presence, not a it's 'point of attachment'.

Any takers?



--
View this message in context: http://freebsd.1045724.x6.nabble.com/Huge-hpet0-interrupt-storm-but-only-if-HGST-HTS721010A9E630-HDD-is-present-at-the-boot-and-formatted-tp6151041.html
Sent from the freebsd-stable mailing list archive at Nabble.com.