Serious boot performance regression on 42167

[stargatefan]

Some change in between 41894 and 42167 is cuasing a serious boot time regression when the system detects a USB hub style card reader. syslog attached.

It was during these events using the onscreen debugger that I was able to see where it was slowing down at.

KERN: usb_disk: device reports a lun count of 1 KERN: usb_disk: vendor_identification "Lexar " KERN: usb_disk: product_identification "JD FireFly " KERN: usb_disk: product_revision_level "1100" KERN: usb_disk: device reports a lun count of 4 KERN: usb_disk: vendor_identification "Generic " KERN: usb_disk: product_identification "USB SD Reader " KERN: usb_disk: product_revision_level "1.00" KERN: usb_disk: vendor_identification "Generic " KERN: usb_disk: product_identification "USB CF Reader " KERN: usb_disk: product_revision_level "1.01" KERN: usb_disk: vendor_identification "Generic " KERN: usb_disk: product_identification "USB SM Reader " KERN: usb_disk: product_revision_level "1.02" KERN: usb_disk: vendor_identification "Generic " KERN: usb_disk: product_identification "USB MS Reader " KERN: usb_disk: product_revision_level "1.03"

[mmlr]

That'd most probably be hrev42058.

[stargatefan]

it took around 10 seconds per line so total boot time bumped to well over 2 minutes, then once it cleared the last line it booted in like 4 seconds. it is a consistent behavior. Sorry for tagging you with the ticket,

[kallisti5]

This is most likely due to the fact that since the card slots are empty, we wait the full 23 second timeout for *each* card slot.

This is a pretty big regression I didn't plan on.

Going to disable my invalid lun delay as empty usb memory sticks are more common then usb cd-roms.

[diver]

Is this changeset in alpha3 branch?

[kallisti5]

nah. I didn't include it in alpha 3 as it wasn't well tested.

[kallisti5]

check hrev42186 and see if it fixes it for you. Wasn't properly handling the NO_MEDIA event. (I hope these card reader lun's give that back when the slot is empty)

[stargatefan]

Replying to kallisti5:

This is most likely due to the fact that since the card slots are empty, we wait the full 23 second timeout for *each* card slot.

This is a pretty big regression I didn't plan on.

Going to disable my invalid lun delay as empty usb memory sticks are more common then usb cd-roms.

At first I thought the build was broke so I did a clean build with jam clean and jam -a to rebuild all targets, after that I started investigating. I will retest with your changes tonight and report back. Maybe it did take more then 10 seconds per slot. at least I know I am not crazy but this card reader has like 6 slots. thats alot of boot slow down for a OS that normally boots faster then I can open a beer !

[stargatefan]

Fixed in 41286

[anevilyak]

Thanks for the update!