Hi, On Wed, Jun 09, 2004 at 03:40:39PM +0200, Gilles Sadowski wrote: > So, what is the final word about this "timeout" problem?
The reason is unknown as of now. It's not that easy as these timeouts may have several reason. Every wrong command that's sent to the scanner can cause timeouts. It doesn't seem to be a general SANE problem. I've tried several scanners on Linux 2.6.7-rc2 with UHCI, OHCI and EHCI USB host controllers and can't reproduce the timeouts. So someone who does see the timeouts needs to investigate the details. E.g. if there is a specific SANE version that does work without trouble. If that's the case, identify the change that causes problems. If it depends on kernel versions, identify the kernel version that started showing those timeouts. I'll try to summarize what I've read here and in the bug tracking system: Timeouts: - HP5370C (avision) on UHCI VIA, Linux 2.6.5 - HP7450C (avision), sane-backends 1.0.14 (also tried latest avision code), Linux 2.6.5 - HP7400 (avision) on UHCI (Intel Corp. 82371AB/EB/MB PIIX4 USB) Asus board, sane-backends 1.0.14, Linux 2.6.3, 2.6.5, 2.6.6 and 2.6.7-rc2-bk7 - HP7450C (avision), there was a hint about the hpusbscsi driver making trouble in that case - Epson Perfection 1670 (snapscan) on UHCI VIA, sane-backends 1.0.14, Linux 2.6.5-mm2 No timeouts: - Several Mustek+Plustek scanners (mustek_usb, plustek, gt68xx), sane-backends 1.0.14, on OHCI + UHCI (VIA PCI card) on Linux 2.6.7-rc2 and Linux 2.4.26 - Avision scanners, on OHCI, sane-backends 1.0.14 + latest avison code on Linux 2.6.3 (UltraSparc +iBook) Typical log: Jun 6 19:06:11 localhost kernel: usb 1-1: bulk timeout on ep3in Jun 6 19:06:11 localhost kernel: usb 1-1: usbfs: USBDEVFS_BULK failed ep 0x83 len 4096 ret -110 An unrelated bug showed the same timeouts: - Epson Perfection 1260 (plustek), sane-backends-1.0.14, on Linux 2.6.7-rc2 and 2.4.25 with scanner.o(!) Reason: the epson backend tries to talk to this scanner There are some other reports about USB trouble but it's unclear if they are related. All these bug reports are newer than 6 weeks. So it's either the new SANE, kernel, libusb release or some other part of software that got changed recently. All but one reports are about the avision backend. Maybe the trouble is located there and the Epson 1670 problem has a different reason? Bye, Henning
