CPU underload

[Eugene Grosbein]

On 07.11.2015 10:57, Adrian Chadd wrote:
> On 1 November 2015 at 23:17, Eugene Grosbein <eugen at grosbein.net> wrote:
>> On 02.11.2015 14:04, Warner Losh wrote:
>>
>>>     Next, USB flash support is not very stable as I have written before,
>>>     lots of USB i/o can easily panic the kernel.
>>>
>>> Does your flash device support trim / delete?
>>
>> Not sure. How do I know? It is just USB pen flash.
> 
> hiya,
> 
> ok, this is next . What's the deal with this? can you panic it and
> report it please? :)

I did earlier in my letter with subject "jemalloc aborts: lloc_arena.c:363: Failed assertion: "p[i] == 0":

> Trap cause = 2 (TLB miss (load or instr. fetch) - kernel mode)
> panic: trap
> Uptime: 3m7s
> panic: bufwrite: buffer is not busy???
> Uptime: 3m7s

However, I cannot reproduce the problem anymore using recent HEAD
compiled without -Os and with kernel having "options USB_HOST_ALIGN=64"
as Warner Losh suggested.