I've tested it on loongson 3aserver mainboard, but I don't have a x86 mainboard, is there anyone test the patch for me on a x86 mainboard, thanks.
>From 2d4b290e9137365d5c35e3b91f68a38ca7dbb059 Mon Sep 17 00:00:00 2001 From: Yi Jun <[email protected]> Date: Tue, 18 Jun 2013 16:31:13 +0800 Subject: [PATCH] Fixed the bug about rs780e gfx port
It is not neccessary to disable gfx port A when there isn't a device, the trainng bit should holded and all lanes should be closed if there isn't a device. Signed-off-by: Yi Jun <[email protected]> --- src/southbridge/amd/rs780/gfx.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/src/southbridge/amd/rs780/gfx.c b/src/southbridge/amd/rs780/gfx.c index 2825925..3a2b466 100644 --- a/src/southbridge/amd/rs780/gfx.c +++ b/src/southbridge/amd/rs780/gfx.c @@ -1123,10 +1123,14 @@ static void dual_port_configuration(device_t nb_dev, device_t dev) /* Link Training was NOT successful */ if (!result) { /* Powers down all lanes for port A */ - /* nbpcie_ind_write_index(nb_dev, 0x65, 0x0f0f); */ + nbpcie_ind_write_index(nb_dev, 0x65, 0x0f0f); + /* prevent from training. */ + set_nbmisc_enable_bits(nb_dev, 0x8, + 0x1 << (dev_ind + 2), + 0x1 << (dev_ind + 2)); /* Note: I have to disable the slot where there isnt a device, * otherwise the system will hang. I dont know why. */ - set_nbmisc_enable_bits(nb_dev, 0x0c, 1 << dev_ind, 1 << dev_ind); + /* set_nbmisc_enable_bits(nb_dev, 0x0c, 1 << dev_ind, 1 << dev_ind); */ } else { /* step 16.b Link Training was successful */ AtiPcieCfg.PortDetect |= 1 << dev_ind; -- 1.8.1.5
-- coreboot mailing list: [email protected] http://www.coreboot.org/mailman/listinfo/coreboot
