Re: [GIT PATCH] TTY/serial driver patches for .40

["Peter Hüwe" <PeterHuewe@gmx.de>]

Am Freitag 27 Mai 2011, 03:00:21 schrieb Greg KH:
> > > Here's the big tty/serial merge for .40
> > 
> > /me confused: this patch
> > 
> > > Felipe Balbi (1):
> > >       tty: make receive_buf() return the amout of bytes received
> > 
> > is more than one year old, which I wouldn't care about, if it didn't
> > introduce a regression on my mackerel
> > (arch/arm/mach-shmobile/board-mackerel.c) board. With it in place
> > printing a lot of text over ssh pauses the output multiple times for
> > several seconds. With this patch reverted the behaviour is back to
> > normal. To reproduce one can either cat a few largish files under /etc
> > or just dmesg - if there have been enough messages accumulated.
> 
> I think I'll be reverting this one.

Hi,

maybe that's related to receive_buf returning an unsigned int, which is 
obviously incompatible with -ENODEV, -EINVAL and the likes, which most of the 
functions now return in case of error.
And unfortunately the return value is unfortunately not checked for errors 
either - so the amount is totally screwed then ;)

See the examples in git
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=b1c43f82c5aa265442f82dba31ce985ebb7aa71c

I could create a patch if desired.

Thanks,
Peter