Hi, Gil Weber <gil.weber@xxxxxxxxxxxxxxxxxxxxxxxxx> writes: >> from your logs below, it seems like you're using atmel_usba_udc.c, >> right? That driver was merged during v3.17 merge window, how come it was >> working for you on v3.12 ? >> >> Can you try v3.17, when that driver was merged ? > > It was already in previous versions as I used it in 3.12 and 3.5. > From what I see in 3.17 it has just been moved to drivers/usb/gadget/udc. > Before that, all udc drivers were in drivers/usb/gadget. oh, you're right. It was just a rename patch. Sorry. > In 3.12 it was working good, I could try to merge my code to intermediate > versions. okay. So one way would be to bisect what happened between v3.12 and v4.4. >> > <7>[ 275.026000] composite_suspend:2122: g_ether gadget: suspend >> > <7>[ 275.141000] composite_suspend:2122: g_ether gadget: suspend >> > ... >> >> what happened during these almost 5 seconds ?? You shouldn't trim log >> like that, there might be important information ;-) > > It was always the same log. > It appears in loop because I removed power detection in atmel_usba_udc.c > driver for testing. Here is a cleaner log with that power detection when > I remove the cable: > > <7>[ 2540.513000] reset_config:626: g_ether gadget: reset config > <7>[ 2540.513000] rndis_disable:616: g_ether gadget: rndis deactivated > <7>[ 2549.289000] composite_suspend:2122: g_ether gadget: suspend > <6>[ 2549.735000] g_ether gadget: high-speed config #2: RNDIS > <7>[ 2549.735000] usba_ep_disable:655: udc: ep_disable: ep3 not enabled > <7>[ 2549.735000] rndis_set_alt:560: g_ether gadget: reset rndis > <7>[ 2549.735000] usba_ep_disable:655: udc: ep_disable: ep1 not enabled > <7>[ 2549.735000] usba_ep_disable:655: udc: ep_disable: ep2 not enabled > <7>[ 2549.735000] rndis_set_alt:565: g_ether gadget: init rndis > <7>[ 2549.735000] rndis_set_alt:593: g_ether gadget: RNDIS RX/TX early activation ... > <7>[ 2549.735000] rndis_open:638: g_ether gadget: rndis_open > <7>[ 2549.735000] rndis_set_param_medium:981: rndis_set_param_medium: 0 4259840 > <7>[ 2549.735000] rndis_set_param_dev:951: rndis_set_param_dev: > <7>[ 2549.736000] rndis_setup:525: g_ether gadget: rndis req21.00 v0000 i0000 l24 > <7>[ 2549.736000] rndis_msg_parser:810: rndis_msg_parser: RNDIS_MSG_INIT > <7>[ 2549.745000] rndis_setup:525: g_ether gadget: rndis req21.00 v0000 i0000 l24 > <7>[ 2549.746000] rndis_msg_parser:810: rndis_msg_parser: RNDIS_MSG_INIT > > After that there is nothing, it just stop. > >> unlikely. RNDIS works fine at least on dwc3 and MUSB (though I haven't >> tested against windows for a few months) >> >> > I cannot say from which version it is not working anymore as the >> > previous version I used was 3.12. >> >> and that was an atmel provided kernel ? Are you, now, using an atmel >> provided kernel or actual mainline ? Does it work against a Linux host ? >> > > I am using mainline kernel. > > I just did a test with a Linux host and it seems to work. > > To be sure, I also test on other Windows hosts and all are having the > issue. interesting. Wonder what Windows is doing differently. Yeah, I guess the best way forward is to find which commit caused the regression and git bisect is the way to do that. Would be nice to see v3.13, v3.14, v3.15... are working or not. The idea is to find the smallest interval for bisection. cheers -- balbi
Attachment:
signature.asc
Description: PGP signature
