Notice: Trying to access array offset on value of type bool in /srv/www/vhosts/openpandora.org/domains/bugs.openpandora.org/httpdocs/scripts/details.php on line 649 FS#131 : bluetooth drivers read timeout too trigger-happy/logic fail?

OpenPandora Main OS

  • Status Closed
  • Percent Complete
    100%
  • Task Type Bug Report
  • Category Core
  • Operating System All
  • Severity Medium
  • Reported Version SuperZaxxon Beta 4
Attached to Project: OpenPandora Main OS
Opened by Urja Rannikko - 27.06.2010
Last edited by Michael Mrozek - 21.07.2010

FS#131 - bluetooth drivers read timeout too trigger-happy/logic fail?

Shamelessly plugged from the forums, to get proper bug tracking and notifications on this. My post: @notaz: + /* there hould be no rx in progress at this point */ + if (info->rx_skb != NULL) { + dev_err(info->dev, "Receive timed out, %i done, %li left, " + "LSR=%02x, IER=%02x\n", info->rx_skb->len + 1, + info->rx_count, hci_h4p_inb(info, UART_LSR), + hci_h4p_inb(info, UART_IER)); + info->rx_timed_out = 1; + } Could you share your rationale for that comment in drivers/bluetooth/hci_h4p/core.c? (Oh and there's a typo, too :P ) EDIT: Asking this because commenting that block out changed status of bluetooth (for me) from unusable (can receive ~20-500kB of data before ppp connection halting) to fully working - used to have big problems playing 32kbps AAC stream (max 2 min), now already 4 mins of 96kbps, but i still think we need some receive timeout mechanism, but this one (as it currently stands) doesnt work. Notaz's reply: I don't really remember, it was more than 1.5 years ago. That driver is a hack, will need to review it, does it work stable with that code removed? My reply: For the non-IRC people: urjaman on IRC said: (22:47:45) urjaman: notaz: about stability: i'm on irc atm with that driver, its been playing internet radio for a few hours without problems now, but i don know for sure. (22:48:00) urjaman: *dont know (22:50:11) urjaman: my logic: if the serial port _never_ misses a byte it should be ok - IMO scary without some receive timeout, but works for me
Closed by  Michael Mrozek
21.07.2010 00:49
Reason for closing:  Fixed
Additional comments about closing:  Tested and working. Will be included in Hotfix 4.
Grazvydas commented on 08.07.2010 12:19
urjaman: is this still stable for you? We are going to do some more tests with ED and if they pass we'll drop that code.
Urja Rannikko commented on 08.07.2010 14:00
Yeah, still stable. No problems with my bluetooth. IMHO it's anyways much better without this code vs. with this code. (Even when I feel that it's a bit unsafe.) It's been on my todo list to add something like a basic kernel timer based timeout (I dont really know linux kernel internals that much, so I've been holding back on the implementation because I've had other stuff to do, and it has worked well without), something that gets pushed to now+1s (or whatever) on start of receiving of a packet, and when it fires the same check (as in this code) would be done.

Loading...

Available keyboard shortcuts

Tasklist

Task Details

Task Editing