Contribute
Register

Haunted by 0xdeadbeef

Status
Not open for further replies.
Joined
Jul 19, 2013
Messages
6
CPU
4700k
Graphics
Intel HD 4600
Mac
Classic Mac
Mobile Phone
Hey folks.

I'm having some trouble with "0xdeadbeef".

As some of you may know, the infamous "0xdeadbeef" is actually a marker used by some Unices to mark regions of memory as safe for allocation. That is, any memory region containing "0xdeadbeefdeadbeefdeadbeef..." is fair game. Any memory not containing "0xdeadbeef" is unsafe to use, because it should only have a different value if had been already allocated; when the memory is freed, it is filled with "0xdeadbeef" again.

My first build was composed of a 4770k processor, 4x8gb of Corsair 1866 MHz memory (cmz32gx3m4x1866c10), a Corsair TX 750 PSU, and a Gigabyte Z87X-OC motherboard.

After some time on Mavericks, perhaps after updating to something ~10.9.2-ish, I started having stability problems. My system would crash about 5 minutes after login. Sometimes it wouldn't boot, giving me a kernel panic with a message like:

panic(cpu 6 caller 0xffff...): "a freed zone element has been modified: expected 0xdeadbeefdeadbeef but found 0xdeadbeef88098809, bits changed 0x56a436e6, at offset 16 of 32 in zone: kalloc.32"@/SourceCache/xnu/xnu-2422.110

I quickly whipped up an Unibeast installer from the latest Mavericks installer, and I couldn't boot from the USB stick without removing a dimm or doing "maxmem=4096". Even after an install, I couldn't boot without less than 32 GiB of RAM.

So I started a new build. I read online in various places that sometimes Gigabyte boards don't like all 4 slots filled. I wanted to downsize my pc, so I opted this time for an Asus mATX board, the ROG Maximus VII Gene. Others here have had some success with it (although it is Z97).

Same problem. I have tried the XMP profile. I have tried relaxing the timings. I have tried lowering the frequency. I have tried upping the DRAM voltage, and VCCSA. No love. I even tried a couple of the n00b overclocking wizards on the Gene. Nothing worked.

Now at this point, you have to say to yourself, it's probably the RAM. Well, I'm currently reproducing the same results with a similar kit of GSkill 1866 MHz memory with the same timings.

The interesting part is the error is always "kalloc." + "32" or "64" + "''@/SourceCache/xnu/xnu-2422.110". Furthermore, it is always "at offset 48 of 64" or "at offset 16 of 32". And I have only had it report erroneous values of "88098809" or "00000000". (Another forum user reported a similar error with "88098809" on the forum).

If I manage to install OS X (ie. by using "maxmem=4096"), and boot with just "-x IGPEnabler=No", then I can make it to login, but the moment after password entry, I get the same type of panic (with no previous stack trace showing, even if I have booted with "-v").

Can anyone shed some light on this? I've seen others on the forums claim their problems are solved with "maxmem=4096", but we can't run on only 1/4 of our memory day to day :'(
 

Going Bald

Moderator
Joined
Aug 14, 2010
Messages
23,225
Motherboard
GA-X58A-UD7-F7
CPU
W3670
Graphics
RX 580
Mac
  1. iMac
  2. MacBook Pro
  3. Mac mini
Classic Mac
  1. iMac
Mobile Phone
  1. Android
Suggest you download and run http://www.memtest.org/ from a USB with all ram slots populated. Run at least 5 or 6 iterations, possible overnight. If you have a bad ram module, this will find it.
 
Joined
Jan 4, 2011
Messages
436
Motherboard
GigaByte Z390 UD Rev 1.1
CPU
i7-9700K
Graphics
RX 5500 XT
Mac
  1. iMac
  2. Mac mini
Mobile Phone
  1. Android
  2. iOS
I too was getting this error on my new build. I ran all memtest programs and all passed. I read a post that it was showing up on HD 4600 only installs. They got past it by using a patch for HD 4600 and another by using a dedicated video card.

I got past it by installing my GTX 60 Ti and no more deadbeef.

I had to install with a -v -x from the installer, but once in and running multibeast 6.4.2. I also used the unibeast 4.0.2 to make the installer for 10.9.4.

Sorry, I can't remember the link to the HD 4600 patch for mavericks.
 
Joined
Jul 19, 2013
Messages
6
CPU
4700k
Graphics
Intel HD 4600
Mac
Classic Mac
Mobile Phone
Suggest you download and run http://www.memtest.org/ from a USB with all ram slots populated. Run at least 5 or 6 iterations, possible overnight. If you have a bad ram module, this will find it.

Hey thanks for the suggestion. Actually I have tested both kits of RAM, with errors only showing up while testing the full Corsair kit at 1333 MHz. At other speeds, or with the GSkill kit, no errors were found.
 
Joined
Jul 19, 2013
Messages
6
CPU
4700k
Graphics
Intel HD 4600
Mac
Classic Mac
Mobile Phone
I read a post that it was showing up on HD 4600 only installs. They got past it by using a patch for HD 4600 and another by using a dedicated video card.

I got past it by installing my GTX 60 Ti and no more deadbeef.

Wow, that's great news.

Sorry, I can't remember the link to the HD 4600 patch for mavericks.

That's ok, I was actually most interested in ruling out hardware failure as a first priority. I don't mind picking up a graphics card now that I have a 4k monitor LOL.

Maybe I'll test that hypothesis, then RMA my Corsair memory, and take back the GSkill since I don't need it. I hope the RMA doesn't take too long... :beachball:
 
Joined
Jul 19, 2013
Messages
6
CPU
4700k
Graphics
Intel HD 4600
Mac
Classic Mac
Mobile Phone
They got past it by using a patch for HD 4600 and another by using a dedicated video card.

I wonder if these people's kernel panics were a result of the same problem I've been having?

I tried rehabman's patch here: http://www.tonymacx86.com/mavericks...-locked-msrs-hp-envy-15-j063cl-i7-4700mq.html

(I just patched the mach_kernel on the Unibeast installer, and confirmed the MD5 matched as expected.)

My kernel panic persists. (Even with "-v -x IGPEnabler=No")

(Currently on: 4770k, GSkill 10/10/11/30 1866 MHz 4x8GiB, Asus Z97 Maximus VII Gene)
 

RehabMan

Moderator
Joined
May 3, 2012
Messages
186,727
Motherboard
Intel DH67BL
CPU
i7-2600K
Graphics
HD 3000
Mac
  1. MacBook Air
Mobile Phone
  1. iOS
I wonder if these people's kernel panics were a result of the same problem I've been having?

I tried rehabman's patch here: http://www.tonymacx86.com/mavericks...-locked-msrs-hp-envy-15-j063cl-i7-4700mq.html

(I just patched the mach_kernel on the Unibeast installer, and confirmed the MD5 matched as expected.)

My kernel panic persists. (Even with "-v -x IGPEnabler=No")

(Currently on: 4770k, GSkill 10/10/11/30 1866 MHz 4x8GiB, Asus Z97 Maximus VII Gene)

Early reboot is unrelated to deadbeef panic. You need to use maxmem=4096 or maxmem=2048 to work around the deadbeef panic for initial installation.

Make sure you created your USB with Unibeast4.
 
Joined
Jan 4, 2011
Messages
436
Motherboard
GigaByte Z390 UD Rev 1.1
CPU
i7-9700K
Graphics
RX 5500 XT
Mac
  1. iMac
  2. Mac mini
Mobile Phone
  1. Android
  2. iOS
Early reboot is unrelated to deadbeef panic. You need to use maxmem=4096 or maxmem=2048 to work around the deadbeef panic for initial installation.

Make sure you created your USB with Unibeast4.

I'll be, I only heard of 4096 or 8192, but not 2048. I went and bought 4gb so I wouldn't have to type this anymore. but I now need maxmem=2048 now.

Very weird.
 
Status
Not open for further replies.
Top