Random panic in IONVMeController #22
Comments
|
This issue seems to be in OSX. People running this ssd in an adapter on real Mac hit the same issue. Waiting for Apple to fix. |
|
Same here, coming back from sleep, or even today suddenly white typing. Thu Sep 5 14:21:33 2019 *** Panic Report *** BSD process name corresponding to current thread: kernel_task Mac OS version: Kernel version: |
|
What model of SSD? |
|
Could be related to overheating issues in the NVMe #49 (comment) |
|
This might also be caused by PowerNap. There are reports of panics when using non-Apple NVMe with PowerNap, especially on Catalina betas. In addition, the panic might also be more likely on Samsung NVMe drives. This might also be connected to heat since some Samsung NVMe drives can get "stuck" in a failed TRIM cycle for minutes at a time during boot up. |
|
It might be connected to dark wake but it can’t be “caused” by dark wake because all it does is trigger wake ups on a timer. Btw I’m assuming you’re referring to dark wake because the “power nap” feature requires special Apple hw support and doesn’t work on any hackintosh. But dark wake is just “power nap” without the power saving features brought by Apple hw. |
I'm not convinced that there is /no/ attempt by macOS to use power savings features on non-Apple hardware (for better or for worse) when PowerNap is enabled. I've read reports of fans behaving differently or not spinning (similar to what real Mac do) on particular hackintoshes during PowerNap dark wakes. But yes, I was referring to dark wake when I said "power nap" above. I apologize for my lack of clarity. |
|
Just following up on this and after putting better cooling on my nvme's some time ago have not had a single issue since. In fact I have put even more cooling in and on full load to the drives the temps hardly move. |
|
That’s great to hear! I’ve read that Samsung drives do run pretty hot on Hackintoshes so it’s great you were able to fix it. |
I bought a HatSSD, clean installed Mojave on it, crashes happen randomly even while using the machine, suddenly kernel panic. |
|
I had the same issue for over two years running the system with a 960EVO samsung ssd.
My bet is that the kernel panic comes from a wrong parameter in mac os that messes up the drive (or the motherboard socked) frying these components over time. The reason I believe that is:
Next month I'll buy another nvme ssd (I don't know if I try another model from samsung - 970?; or from another brand to test it). With it, I'll know if my Motherboard socket is dead (in case it still crashes with the new drive), or if the old drive was the problem (in the case it works fine). |
|
I don't think the bug from Linux is relevant after all: I looked at IONVMeFamily, and haven't found any usage of FLR. Rather, PCI power states are used to reset the controller. |
|
this panic happened on one of units in my care. It's 13 inch model of early 2015 macbook pro. main culprit is the BootRom 188.0.0.0 version introduced in Catalina 10.15.2 which seems to increase power supply to the SSD (not sure what the objective is, but i guess it's performance related) make it run hotter and crashing during IO burst (installing/copying big file/app) doesnt matter what i did, it keep happening, even clean installation of OS will crash in the midle. I only able to install clean OS by plugging my NVME drive to external enclosure. the drive itself is not a culprit here (in contrary with what a lot of people suggest, that drive might have defect). Because i even swap between 3 drives available (with cross brands and capacity) and any of them will crash on this particular macbook, but run just fine on other units. what finally solved issue for me is installing 10.15.3 when it released, forcing the BootRom to go up again to 189.0.0.0 and then combined with downgrading to Mojave and run HFS+ instead of APFS. now it run stable in almost any condition, even during high load. But just for precaution, i also installing mac fan control to monitor and keep the NVME stick cool, since i notice that when it heat up past 50 degree C, the performance seems to drop (stuttering UI, etc). This never happen with other units (i've been installing NVME stick to dozens of macbook pr and air). So, weird indeed, but luckily it solved after much frustration. |
|
We've had a panic on PM981 where we observed the following: I/O Read command (qid=1, opcode=2) timed out. We then got panic log leading to RequestAsyncEvents call. Now, I'm not sure if the panic is due to the I/O command out, or due to the RequestAsyncEvents command timing out, but if we disable RequestAsyncEvents, then we seemingly do not get the panic anymore. It also seems to be safe to do so, as Curiously, We're sometimes getting exactly the same panic on a virtual machine. Coincidentally, AppleNVMeSMARTUserClient does not seem to filter log page IDs, which may cause Apple-specific IDs to hang the controller if a user client does not check controller VID/DID. I only noticed this by accident when I incorrectly assumed the I/O command was submitted to the admin queue. |
|
acidanthera/NVMeFix@66c9de5 should fix this kind of panics. |
|
Thank you for fixing Apple's bugs! |
|
Excuse me for being completely not versed when it comes to Github... How on earth does one use the acidanthera/NVMeFix . I have a late 2015 iMac with a Fledging m13 turbo drive and replaced the normal 3.5 drive with a firecuda. I have reversed operating systems, clean installed, the fusion drive is set up properly. This is the only "fix" I am seeing to the random restart problem and lost and don't know how to use it. More to the question. I have Xcode and have Cloned the code and opened it. How exactly do I put this in the /library/extensions like it says for install. |
|
@ericmschwanke try https://github.com/osy86/HaC-Mini/suites/646220359/artifacts/5470857 I've added an installer option |
unfortunately, i load nvmeFix and lilu, load success, but nothing helps, ssd is 970evo |
|
That’s unfortunate but the issue is outside the scope of this project as it is a third party part that other hackintoshes experience the same problem with. |
|
Just wanted to chime in and say the installer option for NVMe fixup helped me out after I experienced overnight crashes with my Evo 970 & AData NVMe drives (the later used as a raw disk for Virtualbox)... Thank to everyone on this thread for getting that fixed. 👍 |
|
I am still getting this issue. After every few hours, my Macbook shuts down after giving this error. panic(cpu 0 caller 0xffffff7f973c6231): nvme: "Fatal error occurred. CSTS=0xffffffff US[1]=0x0 US[0]=0x20 VID=0x2646 DID=0x5008 Please fix this. |
There is nothing to fix as it’s an Apple side issue. Various workarounds have been proposed here. It none works for you, don’t use nvme. |
|
Hi. Here is the error report. Please help! Thanks. panic(cpu 0 caller 0xffffff7f9d059231): nvme: "Fatal error occurred. CSTS=0xffffffff US[1]=0x0 US[0]=0x26 VID=0x1987 DID=0x5012 BSD process name corresponding to current thread: kernel_task Mac OS version: Kernel version: System uptime in nanoseconds: 100327317615607
|
@deftdawg Did you also experience some random freezes before installing the fix? I'm asking because I get some freezes every now and then, and I got the Evo 970 as well. |
|
Don't remember any random crashes while I was using the machine, rather one day it started regularly crashing overnight when not in use. |
|
@deftdawg @shilch @hostelo @travis-joe having same issues and similar kernel panic reports:
#22 (comment) My personal experience is: SSD read/write speeds are half or less what is expected; system works fine and then when the SSD temperature rises due to load (IO burst, benchmark software, etc.) the performance and responsiveness declines; further testing/benchmarking shows the SSD read/write speeds drop to the single digits. If I reboot or shutdown, on next boot it will load with Kernel Panic report and remain sluggish/poor performance. I must boot and perform an NVram reset to fix the problem. |
|
Hi folks, just to add, I have been experiencing the same issue as OP, documented here: The important point for me is "IONVMeController.cpp:5334" is in common with the OP. Some other panics in this bug refer to a different line in IONVMeController. For me it is always line 5334 in IONVMeController. I use to get this issue more often on old boot roms especially 188.0.0.0.0 but with newer roms it only happens every so often (every 1-3 weeks). I know this is may be out of scope for this project but many people with macs and after market SSDs have this issue (google "IONVMeController.cpp:5334"). I have collected my kernel panic logs here (2019-2021): https://pastebin.com/CxCEmg68
@07151129 was the kernel panic for this fix triggered on line 5334 of IONVMeController.cpp? What is happening around 5334 in IONVMeController.cpp? |
|
Ah, my system just crashed when I was using it. It never happened before. Is this a related issue that could be fixed by enabling NvmeFix? |
me too ！
Process name corresponding to current thread: kernel_task Mac OS version: Kernel version: System uptime in nanoseconds: 7095782912640
|
|
I've just had this error 3 times on Apple Macbook Pro 2015 since yesterday. "panic (...) IONVMeControler.cpp:5320". I'm on Catalina and my disk is PNY CS3030 2TB. The disk temperature is about 36 Celsius and worked OK for few years (Power On Hours: 4112). full kernel panic log:
|
Happens with 1TB/512GB HP EX920. With power nap enabled, device will randomly power down after a few days of dark wakes. Sometimes also happens after a burst of IO (such as a disk scan).
EDIT: Seems to be related to overheating/OSX kernel issues.
The text was updated successfully, but these errors were encountered: