Re: [PATCH] net: Use SK_MEM_QUANTUM as minimum for tcp/udp rmem/wmem

On Wednesday 08/12 at 10:46 -0700, Eric Dumazet wrote:

On Wed, 2015-08-12 at 20:00 +0300, Sorin Dumitru wrote:

quoted

Would clamping the values to a min value, like setsockopt(SO_SNDBUF)
does, be an option?
I still find it odd that SO_SNDBUF limits you, while the /proc
interface doesn't. If you think it's
too much, I'm ok with reverting it since it affects scripts.

On those arches where PAGE_SIZE == 64K(or > 16K) it looks like we have
tcp_wmem[1]
smaller than tcp_wmem[0]. Shouldn't we do something about this?

As long as we do not crash if/when root user changes /proc/sys/net
settings, we are good.

Using "1 1 1" for tcp_{r,w}mem seems not to explode, so this sounds good
to me. I'll send a new patch reverting the original.

Thanks,
Calvin

I would not care if performance is bad if root does something really
stupid. root user is supposed to not mess things just for fun.