#16425 closed bug (fixed)
Printing over a network stalls (regression)
Reported by: | vidrep | Owned by: | axeld |
---|---|---|---|
Priority: | normal | Milestone: | R1/beta3 |
Component: | Network & Internet/TCP | Version: | R1/beta2 |
Keywords: | Cc: | ||
Blocked By: | Blocking: | ||
Platform: | All |
Description
Printer - Epson NX625
Driver - Gutenprint
Protocol - LPR/HP JetDirect
Printing the Haiku test page over a network will stall about 2/3 of the way through the job, and freeze print_server. Alpha 4.1 works. Beta 1 and later do not work.
Attachments (2)
Change History (10)
comment:1 by , 4 years ago
comment:2 by , 4 years ago
Component: | Printing → Network & Internet/TCP |
---|---|
Owner: | changed from | to
by , 4 years ago
Attachment: | capture_hrev51382.pcap added |
---|
by , 4 years ago
Attachment: | capture_current.pcap added |
---|
comment:4 by , 4 years ago
Hi, The issue is easy to see in the pcap. In both cases, we can see that the printer TCP window reaches a size of 0 (meaning the buffer in the printer is full and it cannot accept more packets). On the hrev51382 capture, the computer sends a "zero window probe" after some delay, and does this until the printer sends a "TCP window update" once it has enough space again in its internal buffer. In the new capture, there is no zero window probe, and when the printer sends a window update, the transfer does not resume, and the TCP communication just stops there.
comment:6 by , 4 years ago
Tested with hrev55087 x86x-64. Working again with LPR and HP JetDirect. Closing.
comment:7 by , 4 years ago
Resolution: | → fixed |
---|---|
Status: | new → closed |
After spending 2 days and lots of ink tracking the regression down, I can confirm that hrev51385 is the commit that broke network printing.