summaryrefslogtreecommitdiff
path: root/src/drivers/secunet/dmi
diff options
context:
space:
mode:
authorFurquan Shaikh <furquan@google.com>2020-06-22 10:45:12 -0700
committerFurquan Shaikh <furquan@google.com>2020-06-25 08:08:37 +0000
commit8302585c152926c086107cdf11f48040d0d873d5 (patch)
tree8d1d1aaa91797d16e5e6e47e7831558d0a52b5b6 /src/drivers/secunet/dmi
parent462f3ed111e049cc95e2ee2ce4f4e88a415cc912 (diff)
mb/google/zork: Add support for WiFi power sequencing
This change replaces variant_wifi_romstage_gpio_table() with variant_pcie_power_reset_configure() to handle the reset and power sequencing for WiFi devices pre- and post- v3 version of schematics. These are the requirements that need to be satisfied: 1. As per PCI Express M.2 Specification Revision 3.0, Version 1.2, Section 3.1.4 "Power-up Timing", PERST# should stay disabled until `TPVPGL` time duration after device power has stabilized. Value of TPVPGL is implementation specific. 2. For Intel WiFi chip, it is known to get into a bad state if the above requirement is violated and hence requires a power cycle. 3. On pre-v3 schematics: - For both dalboz and trembyle references, GPIO42 drives WIFI_AUX_RESET_L which is pulled up to PP3300_WIFI. - For both dalboz and trembyle references, PP3300_WIFI is controlled using GPIO29. This pad gets pulled high by default on PWRGOOD because of internal pull-up. But, at RESET# it is known to have a glitch. When GPIO29 gets pulled high, it causes WIFI_AUX_RESET_L to be pulled high as well. This violates the PCIe power sequencing requirements. Hence, for pre-v3 schematics on both dalboz and trembyle, following sequence needs to be followed: a. Assert WIFI_AUX_RESET_L. b. Disable power to WiFi. c. Wait 10ms to allow WiFi power to go low. d. Enable power to WiFi. e. Wait 50ms as per PCIe specification. f. Deassert WIFI_AUX_RESET_L. 4. On v3 schematics: - For trembyle: WIFI_AUX_RESET_L is driven by GPIO86 which has an internal PU as well as an external PU to PP3300_WIFI. - For dalboz: WIFI_AUX_RESET is driven by GPIO29. This is active high and has an internal PU. It also has an external 1K PD to overcome internal PU. - For both dalboz and trembyle references, PP3300_WIFI is controlled by GPIO42 which has an internal PU and external PD. Trembyle schematics have a comment saying strong PD of 2.2K but the stuffed resistor is a weak one (499K). ON dalboz, it uses a weak PD (which doesn't look correct and instead should be a strong PD just like trembyle). Having a strong PD ensures that the WiFi power is kept disabled when coming out of G3 until coreboot configures GPIO42 as high. - Thus, for v3 schematics, following sequence needs to be followed: a. Assert WIFI_AUX_RESET{_L} signal. b. Enable power to WiFi. c. Wait 50ms as per PCIe specification. d. Deassert WIFI_AUX_RESET{_L} signal. BUG=b:157686402, b:158257076 TEST=Verified that QCA and AX200 cards both continue working. Tested QCA on Dalboz and Trembyle. Tested AX200 on morphius. Signed-off-by: Furquan Shaikh <furquan@google.com> Change-Id: I532131ee911d5efb5130d8710f3e01578f6c9627 Reviewed-on: https://review.coreboot.org/c/coreboot/+/42738 Reviewed-by: Aaron Durbin <adurbin@chromium.org> Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Diffstat (limited to 'src/drivers/secunet/dmi')
0 files changed, 0 insertions, 0 deletions