[haiku-bugs] Re: [Haiku] #11020: Boot hangs on rocket

  • From: "mmlr" <trac@xxxxxxxxxxxx>
  • Date: Fri, 02 Jan 2015 12:06:28 -0000

#11020: Boot hangs on rocket
--------------------------+----------------------------
   Reporter:  jstressman  |      Owner:  nobody
       Type:  bug         |     Status:  assigned
   Priority:  normal      |  Milestone:  R1/beta1
  Component:  - General   |    Version:  R1/Development
 Resolution:              |   Keywords:
 Blocked By:              |   Blocking:  7665
Has a Patch:  0           |   Platform:  All
--------------------------+----------------------------
Changes (by mmlr):

 * owner:  mmlr => nobody
 * status:  new => assigned
 * component:  Drivers/USB => - General


Comment:

 The repeat messages stem from the wireless driver probe requests:

 {{{
 KERN: [net/atheroswifi/0] ieee80211_ref_node
 (ieee80211_send_probereq:1748) 0xe2136000<1c:bd:b9:d2:eb:c7> refcnt 3
 KERN: Last message repeated 20 times.
 Last message repeated 2 times
 }}}

 The KERN prefixed ones are from the kernel. The first repeat is the repeat
 indicator of the last kernel log message (the wifi one). The unprefixed
 one is the repeat indicator of the syslog daemon. In this case the kernel
 repeat of 20 is itself repeated twice. In an otherwise idle system this
 does not look problematic.

 As noted above the "/etc/shadow" message  is normal. The shadow password
 file is only created if you actually set a password for your user or
 create other users.

 As a side note: From reading the syslog it looks like you are running your
 disk controller in IDE/compatibility mode which, except for actual
 compatibility with already installed other OSes that require it, is rather
 suboptimal. Using AHCI mode reduces the overhead of disk IO greatly.

 The actual problem is unfortunately still hard to diagnose with the given
 revision. Even though it is only a single hrev, it points to a merge
 commit that brings in a lot of changes at once. Note also that there were
 some issues with the merged scheduler branch early on that only got
 exposed on certain hardware and were subsequently fixed in later hrevs.

 Ideally you could bisect the scheduler branch itself, to find the commit
 within that first caused these symptoms. Unfortunately there are no pre-
 built images of that branch AFAIR.

--
Ticket URL: <https://dev.haiku-os.org/ticket/11020#comment:9>
Haiku <https://dev.haiku-os.org>
Haiku - the operating system.

Other related posts: