user/sven/linux.git/drivers/firmware, branch v5.9.8

firmware: arm_scmi: Move scmi bus init and exit calls into the driver

2020-11-05T10:51:30Z

[ Upstream commit 5a2f0a0bdf201e2183904b6217f9c74774c961a8 ] In preparation to enable building scmi as a single module, let us move the scmi bus {de-,}initialisation call into the driver. The main reason for this is to keep it simple instead of maintaining it as separate modules and dealing with all possible initcall races and deferred probe handling. We can move it as separate modules if needed in future. Link: https://lore.kernel.org/r/20200907195046.56615-3-sudeep.holla@arm.com Tested-by: Cristian Marussi Signed-off-by: Sudeep Holla Signed-off-by: Sasha Levin

firmware: arm_scmi: Fix duplicate workqueue name

2020-11-05T10:51:09Z

[ Upstream commit b9ceca6be43233845be70792be9b5ab315d2e010 ] When more than a single SCMI device are present in the system, the creation of the notification workqueue with the WQ_SYSFS flag will lead to the following sysfs duplicate node warning: sysfs: cannot create duplicate filename '/devices/virtual/workqueue/scmi_notify' CPU: 0 PID: 20 Comm: kworker/0:1 Not tainted 5.9.0-gdf4dd84a3f7d #29 Hardware name: Broadcom STB (Flattened Device Tree) Workqueue: events deferred_probe_work_func Backtrace: show_stack + 0x20/0x24 dump_stack + 0xbc/0xe0 sysfs_warn_dup + 0x70/0x80 sysfs_create_dir_ns + 0x15c/0x1a4 kobject_add_internal + 0x140/0x4d0 kobject_add + 0xc8/0x138 device_add + 0x1dc/0xc20 device_register + 0x24/0x28 workqueue_sysfs_register + 0xe4/0x1f0 alloc_workqueue + 0x448/0x6ac scmi_notification_init + 0x78/0x1dc scmi_probe + 0x268/0x4fc platform_drv_probe + 0x70/0xc8 really_probe + 0x184/0x728 driver_probe_device + 0xa4/0x278 __device_attach_driver + 0xe8/0x148 bus_for_each_drv + 0x108/0x158 __device_attach + 0x190/0x234 device_initial_probe + 0x1c/0x20 bus_probe_device + 0xdc/0xec deferred_probe_work_func + 0xd4/0x11c process_one_work + 0x420/0x8f0 worker_thread + 0x4fc/0x91c kthread + 0x21c/0x22c ret_from_fork + 0x14/0x20 kobject_add_internal failed for scmi_notify with -EEXIST, don't try to register things with the same name in the same directory. arm-scmi brcm_scmi@1: SCMI Notifications - Initialization Failed. arm-scmi brcm_scmi@1: SCMI Notifications NOT available. arm-scmi brcm_scmi@1: SCMI Protocol v1.0 'brcm-scmi:' Firmware version 0x1 Fix this by using dev_name(handle->dev) which guarantees that the name is unique and this also helps correlate which notification workqueue corresponds to which SCMI device instance. Link: https://lore.kernel.org/r/20201014021737.287340-1-f.fainelli@gmail.com Fixes: bd31b249692e ("firmware: arm_scmi: Add notification dispatch and delivery") Signed-off-by: Florian Fainelli [sudeep.holla: trimmed backtrace to remove all unwanted hexcodes and timestamps] Signed-off-by: Sudeep Holla Signed-off-by: Sasha Levin

firmware: arm_scmi: Fix locking in notifications

2020-11-05T10:51:09Z

[ Upstream commit c7821c2d9c0dda0adf2bcf88e79b02a19a430be4 ] When a protocol registers its events, the notification core takes care to rescan the hashtable of pending event handlers and activate all the possibly existent handlers referring to any of the events that are just registered by the new protocol. When a pending handler becomes active the core requests and enables the corresponding events in the SCMI firmware. If, for whatever reason, the enable fails, such invalid event handler must be finally removed and freed. Let us ensure to use the scmi_put_active_handler() helper which handles properly the needed additional locking. Failing to properly acquire all the needed mutexes exposes a race that leads to the following splat being observed: WARNING: CPU: 0 PID: 388 at lib/refcount.c:28 refcount_warn_saturate+0xf8/0x148 Hardware name: ARM LTD ARM Juno Development Platform/ARM Juno Development Platform, BIOS EDK II Jun 30 2020 pstate: 40000005 (nZcv daif -PAN -UAO BTYPE=--) pc : refcount_warn_saturate+0xf8/0x148 lr : refcount_warn_saturate+0xf8/0x148 Call trace: refcount_warn_saturate+0xf8/0x148 scmi_put_handler_unlocked.isra.10+0x204/0x208 scmi_put_handler+0x50/0xa0 scmi_unregister_notifier+0x1bc/0x240 scmi_notify_tester_remove+0x4c/0x68 [dummy_scmi_consumer] scmi_dev_remove+0x54/0x68 device_release_driver_internal+0x114/0x1e8 driver_detach+0x58/0xe8 bus_remove_driver+0x88/0xe0 driver_unregister+0x38/0x68 scmi_driver_unregister+0x1c/0x28 scmi_drv_exit+0x1c/0xae0 [dummy_scmi_consumer] __arm64_sys_delete_module+0x1a4/0x268 el0_svc_common.constprop.3+0x94/0x178 do_el0_svc+0x2c/0x98 el0_sync_handler+0x148/0x1a8 el0_sync+0x158/0x180 Link: https://lore.kernel.org/r/20201013133109.49821-1-cristian.marussi@arm.com Fixes: e7c215f358a35 ("firmware: arm_scmi: Add notification callbacks-registration") Signed-off-by: Cristian Marussi Signed-off-by: Sudeep Holla Signed-off-by: Sasha Levin

firmware: arm_scmi: Add missing Rx size re-initialisation

2020-11-05T10:51:09Z

[ Upstream commit 9724722fde8f9bbd2b87340f00b9300c9284001e ] Few commands provide the list of description partially and require to be called consecutively until all the descriptors are fetched completely. In such cases, we don't release the buffers and reuse them for consecutive transmits. However, currently we don't reset the Rx size which will be set as per the response for the last transmit. This may result in incorrect response size being interpretted as the firmware may repond with size greater than the one set but we read only upto the size set by previous response. Let us reset the receive buffer size to max possible in such cases as we don't know the exact size of the response. Link: https://lore.kernel.org/r/20201012141746.32575-1-sudeep.holla@arm.com Fixes: b6f20ff8bd94 ("firmware: arm_scmi: add common infrastructure and support for base protocol") Reported-by: Etienne Carriere Signed-off-by: Sudeep Holla Signed-off-by: Sasha Levin

firmware: arm_scmi: Expand SMC/HVC message pool to more than one

2020-11-05T10:51:08Z

[ Upstream commit 7adb2c8aaaa6a387af7140e57004beba2c04a4c6 ] SMC/HVC can transmit only one message at the time as the shared memory needs to be protected and the calls are synchronous. However, in order to allow multiple threads to send SCMI messages simultaneously, we need a larger poll of memory. Let us just use value of 20 to keep it in sync mailbox transport implementation. Any other value must work perfectly. Link: https://lore.kernel.org/r/20201008143722.21888-4-etienne.carriere@linaro.org Fixes: 1dc6558062da ("firmware: arm_scmi: Add smc/hvc transport") Cc: Peng Fan Signed-off-by: Etienne Carriere [sudeep.holla: reworded the commit message to indicate the practicality] Signed-off-by: Sudeep Holla Signed-off-by: Sasha Levin

firmware: arm_scmi: Fix ARCH_COLD_RESET

2020-11-05T10:51:08Z

[ Upstream commit 45b9e04d5ba0b043783dfe2b19bb728e712cb32e ] The defination for ARCH_COLD_RESET is wrong. Let us fix it according to the SCMI specification. Link: https://lore.kernel.org/r/20201008143722.21888-5-etienne.carriere@linaro.org Fixes: 95a15d80aa0d ("firmware: arm_scmi: Add RESET protocol in SCMI v2.0") Signed-off-by: Etienne Carriere Signed-off-by: Sudeep Holla Signed-off-by: Sasha Levin

efi/arm64: libstub: Deal gracefully with EFI_RNG_PROTOCOL failure

2020-11-01T11:47:02Z

commit d32de9130f6c79533508e2c7879f18997bfbe2a0 upstream. Currently, on arm64, we abort on any failure from efi_get_random_bytes() other than EFI_NOT_FOUND when it comes to setting the physical seed for KASLR, but ignore such failures when obtaining the seed for virtual KASLR or for early seeding of the kernel's entropy pool via the config table. This is inconsistent, and may lead to unexpected boot failures. So let's permit any failure for the physical seed, and simply report the error code if it does not equal EFI_NOT_FOUND. Cc: # v5.8+ Reported-by: Heinrich Schuchardt Signed-off-by: Ard Biesheuvel Signed-off-by: Greg Kroah-Hartman

firmware: arm_scmi: Fix NULL pointer dereference in mailbox_chan_free

2020-10-29T09:12:02Z

[ Upstream commit 6ed6c558234f0b6c22e47a3c2feddce3d02324dd ] scmi_mailbox is obtained from cinfo->transport_info and the first call to mailbox_chan_free frees the channel and sets cinfo->transport_info to NULL. Care is taken to check for non NULL smbox->chan but smbox can itself be NULL. Fix it by checking for it without which, kernel crashes with below NULL pointer dereference and eventually kernel panic. Unable to handle kernel NULL pointer dereference at virtual address 0000000000000038 Modules linked in: scmi_module(-) Hardware name: ARM LTD ARM Juno Development Platform/ARM Juno Development Platform, BIOS EDK II Sep 2 2020 pstate: 80000005 (Nzcv daif -PAN -UAO BTYPE=--) pc : mailbox_chan_free+0x2c/0x70 [scmi_module] lr : idr_for_each+0x6c/0xf8 Call trace: mailbox_chan_free+0x2c/0x70 [scmi_module] idr_for_each+0x6c/0xf8 scmi_remove+0xa8/0xf0 [scmi_module] platform_drv_remove+0x34/0x58 device_release_driver_internal+0x118/0x1f0 driver_detach+0x58/0xe8 bus_remove_driver+0x64/0xe0 driver_unregister+0x38/0x68 platform_driver_unregister+0x1c/0x28 scmi_driver_exit+0x38/0x44 [scmi_module] ---[ end trace 17bde19f50436de9 ]--- Kernel panic - not syncing: Fatal exception SMP: stopping secondary CPUs Kernel Offset: 0x1d0000 from 0xffff800010000000 PHYS_OFFSET: 0x80000000 CPU features: 0x0240022,25806004 Memory Limit: none ---[ end Kernel panic - not syncing: Fatal exception ]--- Link: https://lore.kernel.org/r/20200908112611.31515-1-sudeep.holla@arm.com Fixes: 5c8a47a5a91d ("firmware: arm_scmi: Make scmi core independent of the transport type") Cc: Cristian Marussi Cc: Viresh Kumar Tested-by: Cristian Marussi Reviewed-by: Cristian Marussi Reviewed-by: Viresh Kumar Signed-off-by: Sudeep Holla Signed-off-by: Sasha Levin

Merge tag 'efi-urgent-for-v5.9-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

2020-09-20T22:18:11Z

Pull EFI fix from Borislav Petkov: "Ensure that the EFI bootloader control module only probes successfully on systems that support the EFI SetVariable runtime service" [ Tag and commit from Ard Biesheuvel, forwarded by Borislav ] * tag 'efi-urgent-for-v5.9-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: efi: efibc: check for efivars write capability

efi: efibc: check for efivars write capability

2020-09-15T15:22:47Z

Branden reports that commit f88814cc2578c1 ("efi/efivars: Expose RT service availability via efivars abstraction") regresses UEFI platforms that implement GetVariable but not SetVariable when booting kernels that have EFIBC (bootloader control) enabled. The reason is that EFIBC is a user of the efivars abstraction, which was updated to permit users that rely only on the read capability, but not on the write capability. EFIBC is in the latter category, so it has to check explicitly whether efivars supports writes. Fixes: f88814cc2578c1 ("efi/efivars: Expose RT service availability via efivars abstraction") Tested-by: Branden Sherrell Link: https://lore.kernel.org/linux-efi/AE217103-C96F-4AFC-8417-83EC11962004@gmail.com/ Signed-off-by: Ard Biesheuvel