Summary: | [2.6.19 regression] ppp over bluetooth fails with "TIOCGSERIAL is not supported" error | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Bill Kenworthy <bill> |
Component: | [OLD] Core system | Assignee: | Gentoo Kernel Bug Wranglers and Kernel Maintainers <kernel> |
Status: | RESOLVED NEEDINFO | ||
Severity: | normal | CC: | liquidx, tester |
Priority: | High | ||
Version: | 2006.1 | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | linux-2.6.19-regression | ||
Package list: | Runtime testing required: | --- | |
Attachments: | debug patch |
Description
Bill Kenworthy
2006-12-28 05:43:30 UTC
Created attachment 108316 [details, diff]
debug patch
please apply this patch and look for new messages in dmesg about rfcomm_dev_add
Hopefully it will clarify things
see comment #1 creating the device succeeds.. I think its a racing problem... pppd probably checks for the device before it gets created... what seems to happen is that the udev event to remove the rfcomm device comes almost immediately after the message to add it.. probably while dund is sleep (for 100ms) to wait for it to appear... what seems to happen is that the udev event to remove the rfcomm device comes almost immediately after the message to add it.. as if the other side was pulling it from us (or if the local kernel does for some reason).. The other side is a Nokia N800. I can provide the result of hcidump if its useful. Olivier, can you open an upstream bug at http://bugzilla.kernel.org with your findings so far? Bill, could you please file a bug report upstream (http://bugzilla.kernel.org) about this? And please post the URL of the upstream bug here, so we can track it. Thanks. Once a bug has been opened upstream please reopen this bug and attach the URL for tracking purposes. |