[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#971686: BUG: Bad page map in process glances + BUG: Bad rss-counter state



Control: tag -1 moreinfo

On Mon, 2020-10-05 at 01:05 +0100, Piotr Drozdek wrote:
> Package: src:linux
> Version: 5.7.10-1~bpo10+1
> Severity: normal
> 
> Dear Maintainer,
> 
> *** Reporter, please consider answering these questions, where appropriate ***
> 
>    * What led up to the situation?
> 
> Running normally for 30 days already
> 
>    * What exactly did you do (or not do) that was effective (or
>      ineffective)?
> 
> Did nothing, running as normal. Process glances is running in the screen
> session and it's providing data to InfluxDB + Grafana. Process glances may or
> may not have it's own bugs, but it should not crash with BUG messages I
> experienced

Indeed it should not.

>    * What was the outcome of this action?
> total glances crash, process had to be restarted by script
> 
>    * What outcome did you expect instead?
> no crash
> 
> *** End of the template - remove these template lines ***
> 
> dmesg:
> [Sat Oct  3 07:18:41 2020] swap_info_get: Bad swap offset entry 3fffffffffeff

This looks like a page table entry that was supposed to be clear, but
has one bit set.  That caused it to be interpreted as a swap entry, and
because swap offsets are stored inverted on x86 it was decoded to an
offset with all but one bit set.

Such a bit flip could be caused by a software bug, but could equally be
a hardware fault.  It's probably worth doing a RAM test (memtest86+)
and checking that the CPU is properly cooled.

> [Sat Oct  3 07:18:41 2020] BUG: Bad page map in process glances  pte:00020000
> pmd:344a58067
> [Sat Oct  3 07:18:41 2020] addr:00007f39fe53b000 vm_flags:08000070
> anon_vma:0000000000000000 mapping:ffff9ab88678bfd8 index:174
> [Sat Oct  3 07:18:41 2020] file:_sfc64.cpython-37m-x86_64-linux-gnu.so
> fault:filemap_fault mmap:btrfs_file_mmap [btrfs] readpage:btrfs_readpage
> [btrfs]
> [Sat Oct  3 07:18:41 2020] CPU: 11 PID: 1342447 Comm: glances Tainted: P
> OE     5.7.0-0.bpo.2-amd64 #1 Debian 5.7.10-1~bpo10+1
[...]

You are using the nvidia proprietary driver, and that could also be to
blame.  Please test whether the issue is reproducible without this
driver.

Ben.

-- 
Ben Hutchings
If you seem to know what you are doing, you'll be given more to do.

Attachment: signature.asc
Description: This is a digitally signed message part


Reply to: