Backup/sp7 porting checkpoints#282
Open
rajaganeshr wants to merge 29 commits into
Open
Conversation
The send_uevent() function was using a devm function to allocate the four uevent environment strings (SOURCE, BUS_NUM, PID, ADDRESS). devm_* allocations are tied to the device's lifetime and are only freed when the device is unregistered. As a result, each time an alert occurred, new memory was allocated but not freed. Replace devm allocator with a regular one, add error checking, and free the memory at end. kobject_uevent_env copies all envp_ext strings into its own internal kobj_uevent_env buffer, so it is fine to free after the call. Tested: Flashed on a Nigeria unit. Monitored the udev event, and verified that the property info match the kernel logs. The allocated kmalloc-192 slabs remained the same before and after the error injection. Signed-off-by: Kyle Nieman <Kyle.Nieman@amd.com>
Backport of upstream commit ea34925. The HID gadget interrupt endpoint bInterval was hard-coded (HS/SS=4 -> 1ms poll), forcing the host xHCI to poll the (always-connected, non-suspendable) iKVM stub keyboard ~1000x/sec over PCIe. That constant traffic prevents the host AMD Data Fabric from reaching deep C-states, capping idle DFCS residency at ~65%. Add an "interval" configfs attribute so create_usbhid.sh can raise the stub keyboard's poll interval (e.g. 16 -> ~4s at HS), drastically cutting BMC->host PCIe wakeups while keeping the anti-suspend stub (preserving the Windows KVM idle fix). Default behavior is unchanged when "interval" is not written. Upstream-Status: Inappropriate Signed-off-by: Ashok Kumar <ashok2ku@amd.com>
Name LTPI alert strap and SPI boot-type GPIOs in AMD platfroms. Use corrected mapping (P0=75, P1=77, P0=117, P1=119) and keep P1 names only on dual-processor platforms. Tested: Verified gpioinfo Signed-off-by: Rajaganesh Rathinasabapathi <rrathina@amd.com>
To avoid JTAG driver hang add 1 sec time out to wait_event_interruptible_timeout() Tested: - verified in Morocco in 24 hr test Signed-off-by: modolaty <mohsen.dolaty@amd.com>
Add support for handeling 2 DPC to SBRMI by reading the DIMM ID from DTS. SP7 only has 1 DPC. but SP8 support 2 DPC. Tested: - tested in Kenya by chnaging Kenya DTS file Signed-off-by: modolaty <mohsen.dolaty@amd.com>
Add DIMM SPD and PMIC devices to SP7 platforms for i3c2 (P0) and i3c3 (P1). The i3c buses will run in i2c mode from BMC side. Tested: - verified in Congo and Morocco Signed-off-by: modolaty <mohsen.dolaty@amd.com>
This reverts commit f5e5fbc. Signed-off-by: Akshay Gupta <akshay.gupta@amd.com>
This reverts commit 868c5c8. Signed-off-by: Akshay Gupta <akshay.gupta@amd.com>
This reverts commit 2914126. Signed-off-by: Akshay Gupta <akshay.gupta@amd.com>
This reverts commit 287ff5e. Signed-off-by: Akshay Gupta <akshay.gupta@amd.com>
Since the static address may be zero, update the device address based on the dynamic address. Signed-off-by: Akshay Gupta <akshay.gupta@amd.com>
Since the static address may be zero, update the device address based on the dynamic address. Signed-off-by: Akshay Gupta <akshay.gupta@amd.com>
- Add device tree entries for Venice A0 to support 2P and 2X1P for RMI and TSI devices - This will provide the dynamic address to be used by user space applicaitons Signed-off-by: Akshay Gupta <akshay.gupta@amd.com>
- Add device tree entries for APML devices (RMI/TSI) for Venice AB/B0 platforms. Signed-off-by: Akshay Gupta <akshay.gupta@amd.com>
- Add I3C device IDs for the Venice AB/B0 and SP8 platforms, for RMI and TSI APML devices. Signed-off-by: Akshay Gupta <akshay.gupta@amd.com>
in SP7 PRB systems: - requested I²C mux driver to disconnect all downstream channels when idle. - this will prevent HPM FRU Read corruption from PCIe cards on i2c-8 Tested: - verified in SP7 PRB systems Signed-off-by: modolaty <mohsen.dolaty@amd.com>
All other SP7 platforms configured the static I3C addresses for SBRMI and SBTSI I3C devices. Without this, P1 SBRMI and SBTSI will fail to bind the driver. ``` i3c_device_match_id: manuf 0x0, part 0x0, inst 0x0, ext_info 0x118 sbtsi_i3c 4-118: SBTSI: PID: 118 sbtsi_i3c 4-118: register sbtsi-9 device APML: Registered SBTSI device at address 0x9 ... i3c_device_match_id: manuf 0x0, part 0x100, inst 0x0, ext_info 0x118 sbtsi_i3c 5-1000118: SBTSI: PID: 1000118 sysfs: cannot create duplicate filename '/class/misc/sbtsi-9' CPU: 2 PID: 30983 Comm: s0-state-mgr Tainted: G O 6.6.92-dirty-9baaae4348bb-g9baaae4348bb #1 Hardware name: AMD Ghana PRB (DT) Call trace: dump_backtrace+0xa0/0x11c show_stack+0x18/0x24 dump_stack_lvl+0x48/0x60 dump_stack+0x18/0x24 sysfs_warn_dup+0x64/0x80 sysfs_do_create_link_sd+0xf0/0xf8 sysfs_create_link+0x20/0x40 device_add+0x2fc/0x7ec device_create_groups_vargs+0xe0/0x134 device_create_with_groups+0x58/0x84 misc_register+0xd8/0x188 create_misc_tsi_device+0x78/0xac sbtsi_i3c_probe+0x140/0x214 i3c_device_probe+0x18/0x24 really_probe+0x160/0x3d8 __driver_probe_device+0x88/0x170 device_driver_attach+0x48/0xac bind_store+0x7c/0xd0 drv_attr_store+0x24/0x38 sysfs_kf_write+0x44/0x54 kernfs_fop_write_iter+0x134/0x1e0 vfs_write+0x190/0x314 ksys_write+0x70/0x108 __arm64_sys_write+0x1c/0x28 invoke_syscall.constprop.0+0x5c/0x100 do_el0_svc+0xa8/0xc8 el0_svc+0x40/0x128 el0t_64_sync_handler+0x120/0x12c el0t_64_sync+0x190/0x194 sbtsi_i3c: probe of 5-1000118 failed with error -17 ``` Statically set the I3C addresses for SBRMI/SBTSI for Ghana as well. Tested: Driver binding for Ghana now succeeds for P1 SBRMI and SBTSI. Signed-off-by: Kyle Nieman <Kyle.Nieman@amd.com>
The `fan_sel_gpio` DTS configuration of `3` didn't properly handle the fan pump controller for Ghana. Add the `fan_sel_gpio` value of 255 for Ghana to indicate there is no fan select as BMC is the only controller for the pump fans. Properly configure PWM0 for P0 pump fans and PWM3 for P1 pump fans. Tested: Verified in Ghana. Could change the pump fan speed for P1. Setting the PWM for P0 no longer causes P1 PWM to be set to 0. Signed-off-by: Kyle Nieman <Kyle.Nieman@amd.com>
Alert_L dts changes added for: - 1P: Eagle & Falcon - 2P: Hornbill, Hornbill-mps and Seagull Signed-off-by: Sathya Priya Kumar <sathyapriya.k@amd.com>
from Aspeed: - jtag: aspeed: Support TCK type with u32 Our JTAG hardware mode can only send a maximum TCK value of 1023, so it needs to be split into multiple steps to support the TCK type with u32. AMD change: - add u8 pad to the struct to put u32 tck on 16-bit boundary Tested: - verified in Morocco Signed-off-by: Billy Tsai <billy_tsai@aspeedtech.com> Signed-off-by: modolaty <mohsen.dolaty@amd.com>
Add new FS1606A driver to SP8 Seagull and Falcon Device Trees Tested: - verified in SP8 Seagull Signed-off-by: modolaty <mohsen.dolaty@amd.com>
Add new System VR FS1606 for SP8 PRBs Falcon and Seagull Tested: - verified in Falcon and Seagull Signed-off-by: modolaty <mohsen.dolaty@amd.com>
Enable MCTP over I3C for SP8 platforms tested: verified on Eagle. Signed-off-by: Mahesh Kurapati <mahesh.kurapati@amd.com>
SP8 PRB has FS1606 as System VRs Received the new driver from Pegatron Adding the new driver Tested: - verified in SP8 Falcon and Seagull Signed-off-by: modolaty <mohsen.dolaty@amd.com>
- Add device tree entries for APML devices (RMI/TSI) for SP8 platforms. Signed-off-by: Akshay Gupta <akshay.gupta@amd.com>
Add DIMM SPD and PMIC devices to SP8 platforms for
i3c2 (P0) and i3c3 (P1).
Tested:
- verified in Seagull
Signed-off-by: Rajaganesh Rathinasabapathi <rrathina@amd.com>
Hornbill MPS device Tree is a copy of Hornbill DTS plus VR for VDDCR_CPU0, VDDCR_CPU1, and VDDIO has been changed to MPS. Tested: - verified in Hornbill MPS Signed-off-by: modolaty <mohsen.dolaty@amd.com>
- Add driver for TDA387XX - DTS binding for TDA387XX on falcon - Minor dts fixes Verified on SP8 Falcon Signed-off-by: Ryan Ilgen <Ryan.Ilgen@amd.com>
Resolve conflict fallout in Hornbill and Seagull DTS by restoring APML node placement and label uniqueness to match integ_sp8 intent. This removes duplicate_label dtc failures while preserving Alert_L and APML topology. Tested: - bitbake obmc-phosphor-image (sp7) Signed-off-by: Rajaganesh Rathinasabapathi <rrathina@amd.com>
rajaganeshr
force-pushed
the
backup/sp7-porting-checkpoints
branch
from
July 17, 2026 16:26
4ed3819 to
b3a8464
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
7 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Sync integ_sp8 with dev-a2