Re: Not enough memory when entering CLI

Joshua,
I have continued troubleshooting the CLI on other machines.
I found out several machines were the CLI works fine but when I exited CLI
the machine hanged.
I removed all memory consumption from the driver and still the machine
hanged when I tried to exit.

What can be the issue? What DEBUG messages should I catch?

Thanks,

Itay

On Sun, Nov 1, 2009 at 4:30 PM, Itay Gazit <[email protected]> wrote:

> Joshua,
>
> Your patch did solve the issue for this specific machine.
>
> Thanks,
>
> Itay
>
>
> On Tue, Oct 27, 2009 at 6:58 PM, Joshua Oreman <[email protected]> wrote:
>
>> On Tue, Oct 27, 2009 at 12:46 PM, Itay Gazit <[email protected]> wrote:
>> > Joshua,
>> >
>> > I agree, the memory map probably have not been initialized by the time I
>> > tried to enter CLI during POST.
>> >
>> > Still I have open question:
>> > How gPXE gain the 16MB memory region? what do I need to change in order
>> to
>> > have 20MB?
>> >
>> > This is not the only machine that has problem entering the CLI. We
>> should
>> > get to the bottom of it. I am willing to help with troubleshooting.
>>
>> The solution is in my previous email:
>>
>> Don't try to enter the CLI during POST. Enter the CLI when gPXE is
>> invoked as a boot device, i.e. when the prompt says "Press Ctrl-B for
>> the gPXE command prompt". If the confusion of having two prompts is
>> too great for you or your users, edit arch/i386/prefix/romprefix.S and
>> change the ROM_BANNER_TIMEOUT line near the top to
>>
>> #define ROM_BANNER_TIMEOUT 0
>>
>> gPXE thinks you have 16MB of memory because, during POST, your BIOS
>> supplies complete junk when gPXE asks it for a memory count. This is
>> not uncommon. Perhaps the POST-time prompt should provide only a
>> limited subset of gPXE's functionality, but that has not yet been
>> implemented. If you insist upon breaking into gPXE before the system
>> is fully initialized, the following patch will cause it to ignore the
>> junk and fall back on int 15,88h, which is a legacy interface that can
>> report up to 64MB of memory; hopefully your BIOS can at least get that
>> right. Again, use DEBUG=memmap to see what it's deciding on.
>>
>> diff --git a/src/arch/i386/firmware/pcbios/memmap.c
>> b/src/arch/i386/firmware/pcbios/memmap.c
>> index 8a30dba..ae9f8f3 100644
>> --- a/src/arch/i386/firmware/pcbios/memmap.c
>> +++ b/src/arch/i386/firmware/pcbios/memmap.c
>> @@ -105,7 +105,7 @@ static unsigned int extmemsize_e801 ( void ) {
>>         * would leave no address space available for 32-bit PCI
>>         * BARs).
>>         */
>> -       if ( extmem == ( 0x400000 - 0x400 ) ) {
>> +       if ( extmem >= ( 0x400000 - 0x400 ) ) {
>>                DBG ( "INT 15,e801 reported whole 4GB; assuming insane\n"
>> );
>>                return 0;
>>        }
>>
>> Let me know if this solves your problem.
>>
>> -- Josh
>>
>
>

------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day 
trial. Simplify your report design, integration and deployment - and focus on 
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july

_______________________________________________
Etherboot-discuss mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/etherboot-discuss