• Open Menu Close Menu
  • Apple
  • Shopping Bag
  • Apple
  • Mac
  • iPad
  • iPhone
  • Watch
  • TV
  • Music
  • Support
  • Search apple.com
  • Shopping Bag

Lists

Open Menu Close Menu
  • Terms and Conditions
  • Lists hosted on this site
  • Email the Postmaster
  • Tips for posting to public mailing lists
Re: shark and kernel/driver profiling
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: shark and kernel/driver profiling


  • Subject: Re: shark and kernel/driver profiling
  • From: Andrew Gallatin <email@hidden>
  • Date: Tue, 27 Feb 2007 15:27:36 -0500 (EST)

Rick Altherr writes:
 >
 > Shark doesn't use the .sym file.  It uses the actual kext on disk
 > along with information about where it was loaded that we retrieve
 > from the running kernel.  However, I have noticed sometimes that

Cool!

 > using the kext search path doesn't work.  The guaranteed way to get
 > kext symbols is to make sure the kext is build with -g and is _not_
 > stripped later.  Then, put it in /System/Library/Extensions and load
 > it using kextload with either the bundle id or the full path in /
 > System/Library/Extensions.  If that doesn't work, then I'd need to
 > see the driver.  Thankfully, I've already seen it once, since I was
 > sent 2 of those cards to run some throughput testing a while back.

I tried and I was not able to make it work, even using -g -O0 built
driver, running from /S/L/E.  According to ktrace, I see it open and
mmap my driver's binary:

<...>
   313 Shark    CALL  lstat(0xb05deaee,0xb05df750)
   313 Shark    NAMI  "/System/Library/Extensions/myri10ge.kext/Contents/MacOS/myri10ge"
   313 Shark    RET   lstat 0
   313 Shark    CALL  access(0xb05df3be,0x4)
   313 Shark    NAMI  "/System/Library/Extensions/myri10ge.kext/Contents/MacOS/myri10ge"
   313 Shark    RET   access 0
   313 Shark    CALL  open(0x5fa8260,0,0)
   313 Shark    NAMI  "/System/Library/Extensions/myri10ge.kext/Contents/MacOS/myri10ge"
   313 Shark    RET   open 9
   313 Shark    CALL  fstat(0x9,0xb05df5f0)
   313 Shark    RET   fstat 0
   313 Shark    CALL  mmap(0,0x144934,0x1,0x1,0x9,0)
   313 Shark    RET   mmap 100683776/0x6005000
   313 Shark    CALL  close(0x9)
   313 Shark    RET   close 0
   313 Shark    CALL  lstat(0xb05deaee,0xb05df750)
   313 Shark    NAMI  "/System/Library/Extensions/IONetworkingFamily.kext/Contents/MacOS/IONetworkingFamily"
<...>

If you're interested in taking this further, I've left the kext at
http://www.myri.com/staff/gallatin/shark/myri10ge.kext.tgz
I ran a "script" of rebuilding the driver, and saved
the logfile here:
http://www.myri.com/staff/gallatin/shark/build.log

 > Also, are you a Leopard seed candidate?

When failing to get symbols, I was running 10.4.8 on a MacPro.

But I do have acccess to seeds. In fact, I'm (happily) running a
trivial dtrace script on 9A303 (9.0.0d4) right now which is showing me
the information I want (mainly because it does not truncate unknown
address to page boundaries).  I realize this is a bit stale, but I
haven't had time to download the latest for a while...

Thanks,
Drew
 _______________________________________________
Do not post admin requests to the list. They will be ignored.
Darwin-kernel mailing list      (email@hidden)
Help/Unsubscribe/Update your Subscription:

This email sent to email@hidden

References: 
 >shark and kernel/driver profiling (From: Andrew Gallatin <email@hidden>)
 >Re: shark and kernel/driver profiling (From: Garth Cummings <email@hidden>)
 >Re: shark and kernel/driver profiling (From: Andrew Gallatin <email@hidden>)
 >Re: shark and kernel/driver profiling (From: Garth Cummings <email@hidden>)
 >Re: shark and kernel/driver profiling (From: Andrew Gallatin <email@hidden>)
 >Re: shark and kernel/driver profiling (From: Garth Cummings <email@hidden>)
 >Re: shark and kernel/driver profiling (From: Andrew Gallatin <email@hidden>)
 >Re: shark and kernel/driver profiling (From: Derek Kumar <email@hidden>)

  • Prev by Date: Re: 64-bit Intel CPU
  • Next by Date: RE: Debugging a Kernel Panic
  • Previous by thread: Re: shark and kernel/driver profiling
  • Next by thread: Re: shark and kernel/driver profiling
  • Index(es):
    • Date
    • Thread