talos-op-linux/drivers/soc/fsl, branch master

soc: fsl: qe: remove set but not used variable 'mm_gc'

2020-01-08T22:02:48+00:00

drivers/soc/fsl/qe/gpio.c: In function qe_pin_request: drivers/soc/fsl/qe/gpio.c:163:26: warning: variable mm_gc set but not used [-Wunused-but-set-variable] commit 1e714e54b5ca ("powerpc: qe_lib-gpio: use gpiochip data pointer") left behind this unused variable. Reported-by: Hulk Robot Signed-off-by: Chen Zhou Signed-off-by: YueHaibing Signed-off-by: Li Yang

soc: fsl: qe: remove PPC32 dependency from CONFIG_QUICC_ENGINE

2019-12-09T19:57:45+00:00

There are also PPC64, ARM and ARM64 based SOCs with a QUICC Engine, and the core QE code as well as net/wan/fsl_ucc_hdlc and tty/serial/ucc_uart has now been modified to not rely on ppcisms. So extend the architectures that can select QUICC_ENGINE, and add the rather modest requirements of OF && HAS_IOMEM. The core code as well as the ucc_uart driver has been tested on an LS1021A (arm), and it has also been tested that the QE code still works on an mpc8309 (ppc). Qiang Zhao has tested that the QE-HDLC code that gets enabled with this works on ARM64. Reviewed-by: Timur Tabi Signed-off-by: Rasmus Villemoes Signed-off-by: Li Yang

soc: fsl: qe: remove unused #include of asm/irq.h from ucc.c

2019-12-09T19:57:45+00:00

When allowing this driver to be built for ARM, the build fails (for CONFIG_SMP=y) since ARM's asm/irq.h header is not self-contained: In file included from drivers/soc/fsl/qe/ucc.c:18:0: >> arch/arm/include/asm/irq.h:34:50: error: unknown type name 'cpumask_t' extern void arch_trigger_cpumask_backtrace(const cpumask_t *mask, But nothing in this file actually uses anything from asm/irq.h - removing this #include generates identical object code, both on PPC32 and on ARM (the latter with a patch added to asm/irq.h to make the build work in the first place). Reported-by: kbuild test robot Signed-off-by: Rasmus Villemoes Signed-off-by: Li Yang

soc: fsl: qe: avoid IS_ERR_VALUE in ucc_fast.c

2019-12-09T19:54:37+00:00

When building this on a 64-bit platform gcc rightly warns that the error checking is broken (-ENOMEM stored in an u32 does not compare greater than (unsigned long)-MAX_ERRNO). Instead, change the ucc_fast_[tr]x_virtual_fifo_base_offset members to s32 and use an ordinary check-for-negative. Also, this avoids treating 0 as "this cannot have been returned from qe_muram_alloc() so don't free it". Reviewed-by: Timur Tabi Signed-off-by: Rasmus Villemoes Signed-off-by: Li Yang

soc: fsl: qe: drop pointless check in qe_sdma_init()

2019-12-09T19:54:36+00:00

The sdma member of struct qe_immap is not at offset zero, so even if qe_immr wasn't initialized yet (i.e. NULL), &qe_immr->sdma would not be NULL. Reviewed-by: Timur Tabi Signed-off-by: Rasmus Villemoes Signed-off-by: Li Yang

soc: fsl: qe: drop use of IS_ERR_VALUE in qe_sdma_init()

2019-12-09T19:54:36+00:00

Now that qe_muram_alloc() returns s32, adapt qe_sdma_init() and avoid another few IS_ERR_VALUE() uses. Reviewed-by: Timur Tabi Signed-off-by: Rasmus Villemoes Signed-off-by: Li Yang

soc: fsl: qe: avoid IS_ERR_VALUE in ucc_slow.c

2019-12-09T19:54:36+00:00

When trying to build this for a 64-bit platform, one gets warnings from using IS_ERR_VALUE on something which is not sizeof(long). Instead, change the various *_offset fields to store a signed integer, and simply check for a negative return from qe_muram_alloc(). Since qe_muram_free() now accepts and ignores a negative argument, we only need to make sure these fields are initialized with -1, and we can just unconditionally call qe_muram_free() in ucc_slow_free(). Note that the error case for us_pram_offset failed to set that field to 0 (which, as noted earlier, is anyway a bogus sentinel value). Reviewed-by: Timur Tabi Signed-off-by: Rasmus Villemoes Signed-off-by: Li Yang

soc: fsl: qe: refactor cpm_muram_alloc_common to prevent BUG on error path

2019-12-09T19:54:36+00:00

If the kmalloc() fails, we try to undo the gen_pool allocation we've just done. Unfortunately, start has already been modified to subtract the GENPOOL_OFFSET bias, so we're freeing something that very likely doesn't exist in the gen_pool, meaning we hit the kernel BUG at lib/genalloc.c:399! Internal error: Oops - BUG: 0 [#1] PREEMPT SMP ARM ... [<803fd0e8>] (gen_pool_free) from [<80426bc8>] (cpm_muram_alloc_common+0xb0/0xc8) [<80426bc8>] (cpm_muram_alloc_common) from [<80426c28>] (cpm_muram_alloc+0x48/0x80) [<80426c28>] (cpm_muram_alloc) from [<80428214>] (ucc_slow_init+0x110/0x4f0) [<80428214>] (ucc_slow_init) from [<8044a718>] (qe_uart_request_port+0x3c/0x1d8) (this was tested by just injecting a random failure by adding "|| (get_random_int()&7) == 0" to the "if (!entry)" condition). Refactor the code so we do the kmalloc() first, meaning that's the thing that needs undoing in case gen_pool_alloc_algo() then fails. This allows a later cleanup to move the locking from the callers into the _common function, keeping the kmalloc() out of the critical region and then, hopefully (if all the muram_alloc callers allow) change it to a GFP_KERNEL allocation. Reviewed-by: Timur Tabi Signed-off-by: Rasmus Villemoes Signed-off-by: Li Yang

soc: fsl: qe: drop broken lazy call of cpm_muram_init()

2019-12-09T19:54:36+00:00

cpm_muram_alloc_common() tries to support a kind of lazy initialization - if the muram_pool has not been created yet, it calls cpm_muram_init(). Now, cpm_muram_alloc_common() is always called under spin_lock_irqsave(&cpm_muram_lock, flags); and cpm_muram_init() does gen_pool_create() (which implies a GFP_KERNEL allocation) and ioremap(), not to mention the fun that ensues from cpm_muram_init() doing spin_lock_init(&cpm_muram_lock); In other words, this has never worked, so nobody can have been relying on it. cpm_muram_init() is called from a subsys_initcall (either from cpm_init() in arch/powerpc/sysdev/cpm_common.c or, via qe_reset(), from qe_init() in drivers/soc/fsl/qe/qe.c). Reviewed-by: Timur Tabi Signed-off-by: Rasmus Villemoes Signed-off-by: Li Yang

soc: fsl: qe: make cpm_muram_free() ignore a negative offset

2019-12-09T19:54:36+00:00

This allows one to simplify callers since they can store a negative value as a sentinel to indicate "this was never allocated" (or store the -ENOMEM from an allocation failure) and then call cpm_muram_free() unconditionally. Reviewed-by: Timur Tabi Signed-off-by: Rasmus Villemoes Signed-off-by: Li Yang