Tianocore: Long time to boot / Menu.

[Jose Trujillo]

Hello All:

I am using the default Tianocore stable (Mr. Chromebox Matt version?) but is taking:
2minutes 30seconds from the appearance of the logo to the Tiano menu / grub bootloader.

After entering the payload there is no more serial debug output.

Anyone had this issue and knows the fix.
Anyone of you know how to enable the serial debug output to Tiano.

Any help will be very much appreciated.

Thank you,
Jose Trujillo.

[Matt DeVillier]

hi Jose,

a long boot time in Tianocore usually means that you have serial output
enabled to a port that doesn't exist. Is this with a debug build? What
board, what serial port config?

When I was first testing this years ago, it took 8 mins to boot on one
board because of this

cheers,
Matt

On Wed, Dec 4, 2019, 8:33 AM Jose Trujillo via coreboot <
[e-mail]> wrote:
> Hello All:
>
> I am using the default Tianocore stable (Mr. Chromebox Matt version?) but
> is taking:
> 2minutes 30seconds from the appearance of the logo to the Tiano menu /
> grub bootloader.
>
> After entering the payload there is no more serial debug output.
>
> Anyone had this issue and knows the fix.
> Anyone of you know how to enable the serial debug output to Tiano.
>
> Any help will be very much appreciated.
>
> Thank you,
> Jose Trujillo.
>
> coreboot mailing list -- [e-mail]
> To unsubscribe send an email to [e-mail]

[Jose Trujillo]

Hello Matt.
I also remember when I was working with a baytrail system I had to attach something to the LPC device in order to prevent this... Once I correctly set the SIO under LPC the problem was gone.

In this case I don't have a board's schematic available to see what is attached but I will start probing to try to find un-initialized hardware.

This board was created by autoport.
It's a Sandybridge based board and what I know is it has a Winbond w83627hg attached to LPC which driver I already added under LPC in devicetree.
But this move didn't fixed this issue this time.

I will choose the Tianocore debug to try to see debug information from it.
Thank you,
Jose Trujillo.
> hi Jose,
>
> a long boot time in Tianocore usually means that you have serial output enabled to a port that doesn't exist. Is this with a debug build? What board, what serial port config?
>
> When I was first testing this years ago, it took 8 mins to boot on one board because of this
>
> cheers,
> Matt
>
> On Wed, Dec 4, 2019, 8:33 AM Jose Trujillo via coreboot  wrote:
> > Hello All:
> >
> > I am using the default Tianocore stable (Mr. Chromebox Matt version?) but is taking:
> > 2minutes 30seconds from the appearance of the logo to the Tiano menu / grub bootloader.
> >
> > After entering the payload there is no more serial debug output.
> >
> > Anyone had this issue and knows the fix.
> > Anyone of you know how to enable the serial debug output to Tiano.
> >
> > Any help will be very much appreciated.
> >
> > Thank you,
> > Jose Trujillo.
> >
> > coreboot mailing list -- [e-mail]
> > To unsubscribe send an email to [e-mail]

[Matt DeVillier]

On Thu, Dec 5, 2019 at 3:24 AM Jose Trujillo  wrote:
> Hello Matt.
> I also remember when I was working with a baytrail system I had to attach something to the LPC device in order to prevent this... Once I correctly set the SIO under LPC the problem was gone.
>
> In this case I don't have a board's schematic available to see what is attached but I will start probing to try to find un-initialized hardware.
>
> This board was created by autoport.
> It's a Sandybridge based board and what I know is it has a Winbond w83627hg attached to LPC which driver I already added under LPC in devicetree.
> But this move didn't fixed this issue this time.

are you getting serial output via it from coreboot? If not, enabling
Tianocore debug output is just going to make that boot time longer.

I'd recommend disabling all serial output and seeing what your boot
time is then. If normal, then you know it's the culprit and can start
debugging the coreboot side of things
> I will choose the Tianocore debug to try to see debug information from it.
> Thank you,
> Jose Trujillo.
>
> hi Jose,
>
> a long boot time in Tianocore usually means that you have serial output enabled to a port that doesn't exist. Is this with a debug build? What board, what serial port config?
>
> When I was first testing this years ago, it took 8 mins to boot on one board because of this
>
> cheers,
> Matt
>
> On Wed, Dec 4, 2019, 8:33 AM Jose Trujillo via coreboot  wrote:
> > Hello All:
> >
> > I am using the default Tianocore stable (Mr. Chromebox Matt version?) but is taking:
> > 2minutes 30seconds from the appearance of the logo to the Tiano menu / grub bootloader.
> >
> > After entering the payload there is no more serial debug output.
> >
> > Anyone had this issue and knows the fix.
> > Anyone of you know how to enable the serial debug output to Tiano.
> >
> > Any help will be very much appreciated.
> >
> > Thank you,
> > Jose Trujillo.
> >
> > coreboot mailing list -- [e-mail]
> > To unsubscribe send an email to [e-mail]

[Jose Trujillo]

Dear Matt/All:

I enabled Tianocore debug in coreboot and the serial debug dump showed me Tianocore was trying to open a ATA / ATAPI device and was getting stuck there, so, i disabled a still driverless "ATA" device devicetree until I attach some driver.

After flashing this change, the first boot/reboot will go without delay (normal fast tianocore boot) but after the second boot/reboot the serial dump shows the following long and repeated loop of the following:

Added variable: 0x0, val size: 0
Found variable: key size: 0x0, val size: 0
Added variable: 0x0, val size: 0
Found variable: key size: 0x0, val size: 0
Added variable: 0x0, val size: 0
Found variable: key size: 0x0, val size: 0
Added variable: 0x0, val size: 0
Found variable: key size: 0x0, val size: 0
Added variable: 0x0, val size: 0
Found variable: key size: 0x0, val size: 0
Added variable: 0x0, val size: 0
Found variable: key size: 0x0, val size: 0

Several minutes later boots normal.
If someone here knows how to fix it or suspect which could be the reason please let me know.
I will ask for help in the EDK2 mail list too.

Thank you,
Jose Trujillo.
> are you getting serial output via it from coreboot? If not, enabling
> Tianocore debug output is just going to make that boot time longer.
>
> I'd recommend disabling all serial output and seeing what your boot
> time is then. If normal, then you know it's the culprit and can start
> debugging the coreboot side of things

[Patrick Georgi]

Am Fr., 6. Dez. 2019 um 17:23 Uhr schrieb Jose Trujillo via coreboot <
[e-mail]>:
> Added variable: 0x0, val size: 0
> Found variable: key size: 0x0, val size: 0
>
> Several minutes later boots normal.
> If someone here knows how to fix it or suspect which could be the reason
> please let me know.
> I will ask for help in the EDK2 mail list too.

You probably have the edk2 side smmstore support added? And maybe not the
very latest coreboot side driver installed (Arthur pushed some important
fixes recently)?

I'm not sure if the edk2 mailing list is willing or able to help here,
given that this is a third-party addition from their point of view.

Patrick

[Jose Trujillo]

Dear Patrick:

I was thinking when I saw the EDK2 maillist that this is only for developers, not support.
Also they kicked us the corebooters out early this year deprecating the coreboot dxe.

I will delay the request to EDK2 maillist until I investigate and add the smmstore support.
I will check the documentation on Monday and try to enable it.

Thank you,
Have an excellent weekend.
Jose Trujillo

‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐
On Friday, December 6, 2019 7:36 PM, Patrick Georgi  wrote:
> Am Fr., 6. Dez. 2019 um 17:23 Uhr schrieb Jose Trujillo via coreboot :
> > Added variable: 0x0, val size: 0
> > Found variable: key size: 0x0, val size: 0
> >
> > Several minutes later boots normal.
> > If someone here knows how to fix it or suspect which could be the reason please let me know.
> > I will ask for help in the EDK2 mail list too.
>
> You probably have the edk2 side smmstore support added? And maybe not the very latest coreboot side driver installed (Arthur pushed some important fixes recently)?
>
> I'm not sure if the edk2 mailing list is willing or able to help here, given that this is a third-party addition from their point of view.
> Patrick
>
> Google Germany GmbH, ABC-Str. 19, 20354 Hamburg
> Registergericht und -nummer: Hamburg, HRB 86891, Sitz der Gesellschaft: Hamburg
> Geschäftsführer: Paul Manicle, Halimah DeLaine Prado

[Matt DeVillier]

are you able to tell what's reading/writing the EFI variables? what
are the few lines right before the loop starts?

On Fri, Dec 6, 2019 at 10:23 AM Jose Trujillo via coreboot
 wrote:
> Dear Matt/All:
>
> I enabled Tianocore debug in coreboot and the serial debug dump showed me Tianocore was trying to open a ATA / ATAPI device and was getting stuck there, so, i disabled a still driverless "ATA" device devicetree until I attach some driver.
>
> After flashing this change, the first boot/reboot will go without delay (normal fast tianocore boot) but after the second boot/reboot the serial dump shows the following long and repeated loop of the following:
>
> Added variable: 0x0, val size: 0
> Found variable: key size: 0x0, val size: 0
> Added variable: 0x0, val size: 0
> Found variable: key size: 0x0, val size: 0
> Added variable: 0x0, val size: 0
> Found variable: key size: 0x0, val size: 0
> Added variable: 0x0, val size: 0
> Found variable: key size: 0x0, val size: 0
> Added variable: 0x0, val size: 0
> Found variable: key size: 0x0, val size: 0
> Added variable: 0x0, val size: 0
> Found variable: key size: 0x0, val size: 0
>
> Several minutes later boots normal.
> If someone here knows how to fix it or suspect which could be the reason please let me know.
> I will ask for help in the EDK2 mail list too.
>
> Thank you,
> Jose Trujillo.
> > are you getting serial output via it from coreboot? If not, enabling
> > Tianocore debug output is just going to make that boot time longer.
> >
> > I'd recommend disabling all serial output and seeing what your boot
> > time is then. If normal, then you know it's the culprit and can start
> > debugging the coreboot side of things
>
> coreboot mailing list -- [e-mail]
> To unsubscribe send an email to [e-mail]

[Jose Trujillo]

Hello All:

I am sorry about the delay.
> are you able to tell what's reading/writing the EFI variables?
> what are the few lines right before the loop starts?

Loading driver at 0x0007FB1B000 EntryPoint=0x0007FB1E34A EmuVariableRuntimeDxei
InstallProtocolInterface: BC62157E-3E33-4FEC-9920-2D3B36D750DF 7F8AC018
ProtectUefiImageCommon - 0x7F8AD640
  - 0x000000007FB1B000 - 0x0000000000004FC0
PROGRESS CODE: V03040002 I0
Initialize buffer from 0x40000 bytes of flash
Found variable: key size: 0x0, val size: 0
Added variable: 0x0, val size: 0
Found variable: key size: 0x0, val size: 0
Added variable: 0x0, val size: 0
Found variable: key size: 0x0, val size: 0
Added variable: 0x0, val size: 0
Found variable: key size: 0x0, val size: 0
Added variable: 0x0, val size: 0
Found variable: key size: 0x0, val size: 0
Added variable: 0x0, val size: 0
Found variable: key size: 0x0, val size: 0
Added variable: 0x0, val size: 0
Found variable: key size: 0x0, val size: 0
Added variable: 0x0, val size: 0
> You probably have the edk2 side smmstore support added?
> And maybe not the very latest coreboot side driver installed (Arthur pushed some important fixes recently)?

SMMSTORE was one of the reasons Patrick, thank you, I am glad to see this has been merged.
Also the Tianocore smmstore patches were added into MrChromebox repository... Thank you Matt.

I am using the default old smmstore driver from coreboot's 4.10 because this project has compilation errors in 4.11.

This problem has been resolved.
3 issues were preventing tianocore from booting correctly:

1.- A driverless ATA device.
2.- SMMSTORE was disabled.
3.- LPC had issues and must be running continously.

Now I got a reliable and fast booting to UEFI with persistent settings.

Thank you for your help.
Jose Trujillo.