Since Linux 4.19 was released, the nvme module is missing

(tomaggio) #1

Since kernel 4.19 i got problems with my nvme SSD samsung 960 evo. The filesystem got corrupted and sometimes the got trace and won’t boot. After a new install i got further kernel Panics at boot (random).
After investigation i found out that the nvme module is missing (nvme and nvme-core). Perhaps the is the source of my problems. Is there another module that replaces nvme or is that a mistake?

(totte) #2

What makes you say that? The modules don’t appear to have been removed.

(tomaggio) #3

nvme-fabrics nvme-fcloop nvmem-rave-sp-eeprom nvmet nvmet-rdma
nvme-fc nvme-loop nvme-rdma nvmet-fc
This are all modules. No nvme or nvme core
When i wright nvme in the modules line /etc/mkinitcpio.conf. The creation of the initram failed.
In the case that module is buildin, how can i figure out that is working?

(totte) #4

Please format your posts when pasting output.

Both modules you ask about are found in Chakra (4.19.12-1):

$ grep -rnw '/usr/lib/modules/4.19.12-1-CHAKRA/modules.builtin' -e 'nvme'

(tomaggio) #5

Sorry Totte for that fast shot.


(totte) #6

Sorry, what? Is the issue resolved?

(tomaggio) #7

No it’s not solved but i have no clue to start further investigation.
In the moment the pc is ok an i have to wait until it’s reappears.

(totte) #8

If these issues happen somewhat randomly, could you read the Arch Linux wiki article on troubleshooting, and see about getting information from the boot process and kernel panics? I’m specifically suggesting that you increase log verbosity, and make sure that the logs are saved. Also see this tip:

Pass the kernel parameter oops=panic at boot or write 1 to /proc/sys/kernel/panic_on_oops to force a recoverable oops to issue a panic instead. This is advisable if you are concerned about the small chance of system instability resulting from an oops recovery which may make future errors difficult to diagnose.

(tomaggio) #9

Thanks for your help Totte;
I appreciate it

(Luca Giambonini) #10

There are some fixes in kernel 4.19.19 related to nvme.

nvmet-rdma: fix null dereference under heavy load

commit 5cbab6303b4791a3e6713dfe2c5fda6a867f9adc upstream.

Under heavy load if we don't have any pre-allocated rsps left, we
dynamically allocate a rsp, but we are not actually allocating memory
for nvme_completion (rsp->req.rsp). In such a case, accessing pointer
fields (req->rsp->status) in nvmet_req_init() will result in crash.