Date: Fri, 12 Oct 2012 17:04:24 -0700 From: Garrett Cooper <yanegomi@gmail.com> To: d@delphij.net Cc: freebsd-current@freebsd.org Subject: Re: boot2/loader: serial port handling Message-ID: <CAGH67wQNr%2BZtZPUs-NfLYixEDtzv_O6q4kFhcjFyfacMPfssBw@mail.gmail.com> In-Reply-To: <5078A803.7070705@delphij.net> References: <5078A803.7070705@delphij.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, Oct 12, 2012 at 4:30 PM, Xin Li <delphij@delphij.net> wrote:
> Hi,
>
> We have some rather hacky quick hack at $WORK that addresses a problem
> we have found with boot2/loader and wants to share it and see if we can
> have more neat solution.
>
> Here is the problem: the current boot2 and loader have various places
> where it considers serial port to exist. When the port is not there,
> the code would hang because it tests whether the hardware's -READY bit,
> basically something like:
>
> do {
> } while (inpb(state register) & READY_BIT);
>
> This unfortunately would enter an infinite loop when the device is not
> present -- all in operations would get all bits set.
>
> To reproduce this, one can compile boot2/loader with non-existent port
> and the system will hang at very early stage of boot.
>
> ---
>
> Because boot2 is size constrained we can not use very sophisticated
> detection logic there, what I did is to use something like:
>
> outb(line control register, word)
> if (inb(line control register) != word)
> Disable the serial port read/write
>
> For loader I'm not sure if we should use better detection logic. By the
> way, it seems that the system may force using the default console in
> loader regardless if the detection logic said no, if it decides that's
> the only usable one.
>
> So what would be the right way to solve these issue?
Have you tried out Andriy's commit yet to loader(8) (r241301)?
Cheers,
-Garrett
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAGH67wQNr%2BZtZPUs-NfLYixEDtzv_O6q4kFhcjFyfacMPfssBw>