Re: [PATCH v5 2/5] USB: use genalloc for USB HCs with local memory
From: Laurentiu Tudor
Date: Wed May 22 2019 - 06:48:28 EST
Hello Alan,
On 21.05.2019 20:20, Alan Stern wrote:
> On Tue, 21 May 2019 laurentiu.tudor@xxxxxxx wrote:
>
>> From: Laurentiu Tudor <laurentiu.tudor@xxxxxxx>
>>
>> For HCs that have local memory, replace the current DMA API usage
>> with a genalloc generic allocator to manage the mappings for these
>> devices. To help users, introduce a new HCD API,
>> usb_hcd_setup_local_mem() that will setup up the genalloc backing
>> up the device local memory. It will be used in subsequent patches.
>> This is in preparation for dropping the existing "coherent" dma
>> mem declaration APIs. Current implementation was relying on a short
>> circuit in the DMA API that in the end, was acting as an allocator
>> for these type of devices.
>>
>> For context, see thread here: https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Flkml.org%2Flkml%2F2019%2F4%2F22%2F357&data=02%7C01%7Claurentiu.tudor%40nxp.com%7Cdc4a76b04b874b75bd5408d6de109d86%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C636940560281957207&sdata=JQ1iJc%2Fw%2BwHdTFvaOT%2FYHBx%2BVQlpHWlf346q7sclfcg%3D&reserved=0
>>
>> Signed-off-by: Laurentiu Tudor <laurentiu.tudor@xxxxxxx>
>> Signed-off-by: Fredrik Noring <noring@xxxxxxxxxx>
>> ---
>
>> --- a/drivers/usb/core/hcd.c
>> +++ b/drivers/usb/core/hcd.c
>> @@ -29,6 +29,8 @@
>> #include <linux/workqueue.h>
>> #include <linux/pm_runtime.h>
>> #include <linux/types.h>
>> +#include <linux/genalloc.h>
>> +#include <linux/io.h>
>>
>> #include <linux/phy/phy.h>
>> #include <linux/usb.h>
>> @@ -3039,6 +3041,40 @@ usb_hcd_platform_shutdown(struct platform_device *dev)
>> }
>> EXPORT_SYMBOL_GPL(usb_hcd_platform_shutdown);
>>
>> +int usb_hcd_setup_local_mem(struct usb_hcd *hcd, phys_addr_t phys_addr,
>> + dma_addr_t dma, size_t size)
>> +{
>> + int err;
>> + void __iomem *local_mem;
>> +
>> + hcd->localmem_pool = devm_gen_pool_create(hcd->self.sysdev, PAGE_SHIFT,
>> + dev_to_node(hcd->self.sysdev),
>> + "ohci-hcd");
>
> Surely that string is a mistake. You could use
> dev_name(hcd->self.sysdev) or a name passed by the caller.
Sorry, missed that. Will go with dev_name() if that's ok with you.
>> + if (IS_ERR(hcd->localmem_pool))
>> + return PTR_ERR(hcd->localmem_pool);
>> +
>> + local_mem = devm_memremap(hcd->self.sysdev, phys_addr,
>> + size, MEMREMAP_WC);
>> + if (!local_mem)
>> + return -ENOMEM;
>> +
>> + /*
>> + * Here we pass a dma_addr_t but the arg type is a phys_addr_t.
>> + * It's not backed by system memory and thus there's no kernel mapping
>> + * for it.
>> + */
>> + err = gen_pool_add_virt(hcd->localmem_pool, (unsigned long)local_mem,
>> + dma, size, dev_to_node(hcd->self.sysdev));
>> + if (err < 0) {
>> + dev_err(hcd->self.sysdev, "gen_pool_add_virt failed with %d\n",
>> + err);
>> + return err;
>> + }
>> +
>> + return 0;
>> +}
>> +EXPORT_SYMBOL_GPL(usb_hcd_setup_local_mem);
>
> If you have a usb_hcd_setup_local_mem() function then you should also
> have a usb_hcd_remove_local_mem() function.
Even if all resources that are allocated are device managed?
>> --- a/drivers/usb/host/ohci-mem.c
>> +++ b/drivers/usb/host/ohci-mem.c
>> @@ -36,6 +36,12 @@ static void ohci_hcd_init (struct ohci_hcd *ohci)
>>
>> static int ohci_mem_init (struct ohci_hcd *ohci)
>> {
>> + if (ohci_to_hcd(ohci)->localmem_pool) {
>> + ohci->td_cache = NULL;
>> + ohci->ed_cache = NULL;
>> + return 0;
>> + }
>
> This really isn't necessary. The entire ohci_hcd structure is
> initialized to 0 when it is first allocated.
Will drop.
>> --- a/include/linux/usb/hcd.h
>> +++ b/include/linux/usb/hcd.h
>> @@ -216,6 +216,9 @@ struct usb_hcd {
>> #define HC_IS_RUNNING(state) ((state) & __ACTIVE)
>> #define HC_IS_SUSPENDED(state) ((state) & __SUSPEND)
>>
>> + /* allocator for HCs having local memory */
>
> "allocator" is the wrong word -- an allocator is something that
> allocates. Perhaps "memory area" or something along those lines.
Alright, I will rework.
---
Thanks & Best Regards, Laurentiu