Hi,
I am likely unable to work on this until after the 14th. (with final exams finishing this week) As I understand it, unfixed boards will be disabled from automatic building after the 14th, but will not be dropped until the 31st?
-Matt
On Mon, Dec 9, 2019 at 10:55 AM Michal Zygowski michal.zygowski@3mdeb.com wrote:
On 09.12.2019 05:36, Matt B wrote:
Hi,
Hi Matt,
Thanks for the pointer.
I need a bit more context here, being completely unfamiliar with how coreboot works. I've never done any non-userspace programming, and this is the first time I've gotten a freshly-compiled coreboot to actually boot. I do have most of a degree in computer engineering, but that's unfortunately 99% theory with little-to-no hands-on experience.
A) If I understand correctly, [1] is the complete change (not the last commit in a series for this board) to remove romcc usage from asrock/imb-a180. The same changes need to be applied to asus/am1i-a.
B) If I understand correctly, the necessary family16h changes have been made elsewhere, so only changes to the mainboard-specific code need to be made. This would explain why there is a net removal of code, not addition.
C) I can't say that I really understand the contents of this file. Can someone explain how
/* Set LPC decode enables. */ pci_devfn_t dev = PCI_DEV(0, 0x14, 3); pci_write_config32(dev, 0x44, 0xff03ffd5);
and
/* Enable the AcpiMmio space */ outb(0x24, 0xcd6); outb(0x1, 0xcd7);
and
/* Disable PCI-PCI bridge and release GPIO32/33 for other uses. */ outb(0xea, 0xcd6); outb(0x1, 0xcd7);
becomes
/* Disable PCI-PCI bridge and release GPIO32/33 for other uses. */ pm_write8(0xea, 0x1);
?
We did a lot of cleanup with Kyösti and unified implementations of LPC decode enables and ACPIMMIO. Those instructions you point to were moved to southbridge bootblock initialization, because it is not mainboard specific. Also pm_write8(0xea, 0x1); does the same thing as outb(0xea, 0xcd6); and outb(0x1, 0xcd7); but in more readable form. See: https://review.coreboot.org/c/coreboot/+/37177 https://review.coreboot.org/c/coreboot/+/37329/ https://review.coreboot.org/c/coreboot/+/37168/
D) I'm a bit confused about what [2] is. Is this an earlier, unrefined version of the changes?
I don't think that a naive transform would be wise, as there is a lot more stuff going on in [3] :P For instance:
- would
/* In Hudson RRG, PMIOxD2[5:4] is "Drive strength control for
- LpcClk[1:0]". To be consistent with Parmer, setting to 4mA
- even though the register is not documented in the Kabini BKDG.
- Otherwise the serial output is bad code.
*/ outb(0xD2, 0xcd6); outb(0x00, 0xcd7);
and
This was also moved to southbridge bootblock initialization - not mainboard specific.
/* Disable PCI-PCI bridge and release GPIO32/33 for other uses. */
outb(0xEA, 0xcd6); outb(0x1, 0xcd7);
be transformed similarly to how
/* Disable PCI-PCI bridge and release GPIO32/33 for other uses. */ outb(0xea, 0xcd6); outb(0x1, 0xcd7);
coreboot uses lowercase for hex values. This should be transformed to pm_write8(0xea, 1).
was? 2) would
/* Set LPC decode enables. */ pci_devfn_t dev2 = PCI_DEV(0, 0x14, 3); pci_write_config32(dev2, 0x44, 0xff03ffd5);
and
/* Enable the AcpiMmio space */ outb(0x24, 0xcd6); outb(0x1, 0xcd7);
completely disappear? what about
hudson_lpc_port80();
in the middle? 4) Would
/* Configure ClkDrvStr1 settings */ addr32 = (u32 *)0xfed80e24; *addr32 = 0x030800aa;
/* Configure MiscClkCntl1 settings */
addr32 = (u32 *)0xfed80e40; *addr32 = 0x000c4050; /* enable SIO LPC decode */ dev = PCI_DEV(0, 0x14, 3); byte = pci_read_config8(dev, 0x48); byte |= 3; /* 2e, 2f & 4e, 4f */ pci_write_config8(dev, 0x48, byte); ite_gpio_conf(GPIO_DEV); ite_evc_conf(ENVC_DEV); ite_conf_clkin(CLKIN_DEV, ITE_UART_CLK_PREDIVIDE_48); ite_enable_serial(SERIAL_DEV, CONFIG_TTYS0_BASE); ite_kill_watchdog(GPIO_DEV); /*
- On Larne, after LpcClkDrvSth is set, it needs some time to be stable,
- because of the buffer ICS551M
*/ for (i = 0; i < 200000; i++) val = inb(0xcd6);
remain unchanged?
It should be something like:
/* Configure ClkDrvStr1 settings */ misc_write32(0x24, 0x030800aa); /* Configure MiscClkCntl1 settings */ misc_write32(0x40, 0x000c4050); ite_gpio_conf(GPIO_DEV); ite_evc_conf(ENVC_DEV); ite_conf_clkin(CLKIN_DEV, ITE_UART_CLK_PREDIVIDE_48); ite_enable_serial(SERIAL_DEV, CONFIG_TTYS0_BASE); ite_kill_watchdog(GPIO_DEV); /*
- On Larne, after LpcClkDrvSth is set, it needs some time to be stable,
- because of the buffer ICS551M
*/ for (i = 0; i < 200000; i++) val = inb(0xcd6);
The rest is already done in southbridge bootblock init. The file with modifications should include #include <amdblocks/acpimmio.h>
Regards,
-- Michał Żygowski Firmware Engineerhttp://3mdeb.com | @3mdeb_com
Sincerely, -Matt
[1] https://review.coreboot.org/c/coreboot/+/37453/9 [2] https://review.coreboot.org/c/coreboot/+/37440/10 [3] src/mainboard/asus/am1i-a/romstage.c (for reference: https://pastebin.com/kSka2YL7)
On Sun, Dec 8, 2019 at 6:57 PM Kyösti Mälkki kyosti.malkki@gmail.com wrote:
On Mon, Dec 9, 2019 at 12:32 AM Matt B matthewwbradley6@gmail.com wrote:
Question specifically for Kyösti Mälkki or those similarly
knowledgeable:
My current .config seems to say I'm still using romcc.
Please read the recent thread "AMD AGESA board removals".
I assume I've pulled the latest ("git clone
https://review.coreboot.org/coreboot" as of a couple days ago) so would those changes you mentioned be present?
If so, is there an option to test the C bootblock in the menu somewhere?
There will be no option as romcc will be gone for good. You should do a checkout on commit f66da11 [1], make similar changes to asus/am1i-a, and push your work for review. I have done quick boottest on asrock/imb-a180 with commit ca150dc [2] and it got past the made bootblock changes.
[1] https://review.coreboot.org/c/coreboot/+/37453/9 [2] https://review.coreboot.org/c/coreboot/+/37440/10
Kyösti
coreboot mailing list -- coreboot@coreboot.org To unsubscribe send an email to coreboot-leave@coreboot.org
coreboot mailing list -- coreboot@coreboot.org To unsubscribe send an email to coreboot-leave@coreboot.org