next take at setting up a dma mask by default for platform devices
Subsystems (17)
Hi all,
this is another attempt to make sure the dma_mask pointer is always initialized for platform devices. Not doing so lead to lots of boilerplate code, and makes platform devices different from all our major busses like PCI where we always set up a dma_mask. In the long run this should also help to eventually make dma_mask a scalar value instead of a pointer and remove even more cruft.
The bigger blocker for this last time was the fact that the usb subsystem uses the presence or lack of a dma_mask to check if the core should do dma mapping for the driver, which is highly unusual. So we fix this first. Note that this has some overlap with the pending desire to use the proper dma_mmap_coherent helper for mapping usb buffers. The first two patches from this series should probably go into 5.3 and then uses as the basis for the decision to use dma_mmap_coherent.
PATCH 1/6 usb: don't create dma pools for HCDs with a localmem_pool
If the HCD provides a localmem pool we will never use the DMA pools, so don't create them.
--- drivers/usb/core/buffer.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
drivers/usb/core/buffer.c +3 -3
@@ -66,9 +66,9 @@ int hcd_buffer_create(struct usb_hcd *hcd)char name[16 ];int i, size; -if (!IS_ENABLED (CONFIG_HAS_DMA ) || - (!is_device_dma_capable (hcd->self.sysdev) && - !hcd->localmem_pool)) +if (hcd->localmem_pool || + !IS_ENABLED (CONFIG_HAS_DMA ) || + !is_device_dma_capable (hcd->self.sysdev))return 0 ;for (i =0 ; i <HCD_BUFFER_POOLS ; i++) {
Signature · 1 line
2.20.1
PATCH 2/6 usb: add a hcd_uses_dma helper
In reply to Christoph HellwigThe USB buffer allocation code is the only place in the usb core (and in fact the whole kernel) that uses is_device_dma_capable, while the URB mapping code uses the uses_dma flag in struct usb_bus. Switch the buffer allocation to use the uses_dma flag used by the rest of the USB code, and create a helper in hcd.h that checks this flag as well as the CONFIG_HAS_DMA to simplify the caller a bit.
--- drivers/usb/core/buffer.c | 10 +++------- drivers/usb/core/hcd.c | 4 ++-- drivers/usb/dwc2/hcd.c | 2 +- include/linux/usb.h | 2 +- include/linux/usb/hcd.h | 3 +++ 5 files changed, 10 insertions(+), 11 deletions(-)
drivers/usb/core/buffer.c +3 -7
@@ -66,9 +66,7 @@ int hcd_buffer_create(struct usb_hcd *hcd)char name[16 ];int i, size; -if (hcd->localmem_pool || - !IS_ENABLED (CONFIG_HAS_DMA ) || - !is_device_dma_capable (hcd->self.sysdev)) +if (hcd->localmem_pool || !hcd_uses_dma (hcd))return 0 ;for (i =0 ; i <HCD_BUFFER_POOLS ; i++) {
@@ -129,8 +127,7 @@ void *hcd_buffer_alloc(return gen_pool_dma_alloc (hcd->localmem_pool, size, dma);/* some USB hosts just use PIO */ -if (!IS_ENABLED (CONFIG_HAS_DMA ) || - !is_device_dma_capable (bus->sysdev)) { +if (!hcd_uses_dma (hcd)) { *dma = ~(dma_addr_t )0 ;return kmalloc (size, mem_flags); }
@@ -160,8 +157,7 @@ void hcd_buffer_free(return ; } -if (!IS_ENABLED (CONFIG_HAS_DMA ) || - !is_device_dma_capable (bus->sysdev)) { +if (!hcd_uses_dma (hcd)) {kfree (addr);return ; }
drivers/usb/core/hcd.c +2 -2
@@ -1412,7 +1412,7 @@ int usb_hcd_map_urb_for_dma(struct usb_hcd *hcd, struct urb *urb,if (usb_endpoint_xfer_control (&urb->ep->desc)) {if (hcd->self.uses_pio_for_control)return ret; -if (IS_ENABLED (CONFIG_HAS_DMA ) && hcd->self.uses_dma) { +if (hcd_uses_dma (hcd)) {if (is_vmalloc_addr (urb->setup_packet)) {WARN_ONCE (1 ,"setup packet is not dma capable\n" );return -EAGAIN ;
@@ -1446,7 +1446,7 @@ int usb_hcd_map_urb_for_dma(struct usb_hcd *hcd, struct urb *urb, dir =usb_urb_dir_in (urb) ?DMA_FROM_DEVICE :DMA_TO_DEVICE ;if (urb->transfer_buffer_length !=0 && !(urb->transfer_flags &URB_NO_TRANSFER_DMA_MAP )) { -if (IS_ENABLED (CONFIG_HAS_DMA ) && hcd->self.uses_dma) { +if (hcd_uses_dma (hcd)) {if (urb->num_sgs) {int n;
drivers/usb/dwc2/hcd.c +1 -1
@@ -4608,7 +4608,7 @@ static int _dwc2_hcd_urb_enqueue(struct usb_hcd *hcd, struct urb *urb, buf = urb->transfer_buffer; -if (hcd->self.uses_dma) { +if (hcd_uses_dma (hcd)) {if (!buf && (urb->transfer_dma &3 )) { dev_err(hsotg->dev, "%s: unaligned transfer with no transfer_buffer",
include/linux/usb.h +1 -1
@@ -1457,7 +1457,7 @@ typedef void (*usb_complete_t)(struct urb *);* field rather than determining a dma address themselves. * * Note that transfer_buffer must still be set if the controller -* does not support DMA (as indicated by bus.uses_dma) and when talking +* does not support DMA (as indicated by hcd_uses_dma()) and when talking * to root hub. If you have to trasfer between highmem zone and the device * on such controller, create a bounce buffer or bail out with an error. * If transfer_buffer cannot be set (is in highmem) and the controller is DMA
include/linux/usb/hcd.h +3 -0
@@ -422,6 +422,9 @@ static inline bool hcd_periodic_completion_in_progress(struct usb_hcd *hcd,return hcd->high_prio_bh.completing_ep == ep; } +#define hcd_uses_dma (hcd) \ + (IS_ENABLED(CONFIG_HAS_DMA) && (hcd)->self.uses_dma) +extern int usb_hcd_link_urb_to_ep (struct usb_hcd *hcd,struct urb *urb);extern int usb_hcd_check_unlink_urb (struct usb_hcd *hcd,struct urb *urb,int status);
Signature · 1 line
2.20.1
PATCH 3/6 usb: add a HCD_DMA flag instead of guestimating DMA capabilities
In reply to Christoph HellwigThe usb core is the only major place in the kernel that checks for a non-NULL device dma_mask to see if a device is DMA capable. This is generally a bad idea, as all major busses always set up a DMA mask, even if the device is not DMA capable - in fact bus layers like PCI can't even know if a device is DMA capable at enumeration time. This leads to lots of workaround in HCD drivers, and also prevented us from setting up a DMA mask for platform devices by default last time we tried.
Replace this guess with an explicit HCD_DMA that is set by drivers that appear to have DMA support.
--- drivers/staging/octeon-usb/octeon-hcd.c | 2 +- drivers/usb/core/hcd.c | 1 - drivers/usb/dwc2/hcd.c | 6 +++--- drivers/usb/host/ehci-grlib.c | 2 +- drivers/usb/host/ehci-hcd.c | 2 +- drivers/usb/host/ehci-pmcmsp.c | 2 +- drivers/usb/host/ehci-ppc-of.c | 2 +- drivers/usb/host/ehci-ps3.c | 2 +- drivers/usb/host/ehci-sh.c | 2 +- drivers/usb/host/ehci-xilinx-of.c | 2 +- drivers/usb/host/fhci-hcd.c | 2 +- drivers/usb/host/fotg210-hcd.c | 2 +- drivers/usb/host/imx21-hcd.c | 2 +- drivers/usb/host/isp116x-hcd.c | 6 ------ drivers/usb/host/isp1362-hcd.c | 5 ----- drivers/usb/host/ohci-hcd.c | 2 +- drivers/usb/host/ohci-ppc-of.c | 2 +- drivers/usb/host/ohci-ps3.c | 2 +- drivers/usb/host/ohci-sa1111.c | 2 +- drivers/usb/host/ohci-sm501.c | 2 +- drivers/usb/host/ohci-tmio.c | 2 +- drivers/usb/host/oxu210hp-hcd.c | 3 --- drivers/usb/host/r8a66597-hcd.c | 6 ------ drivers/usb/host/sl811-hcd.c | 6 ------ drivers/usb/host/u132-hcd.c | 2 -- drivers/usb/host/uhci-grlib.c | 2 +- drivers/usb/host/uhci-pci.c | 2 +- drivers/usb/host/uhci-platform.c | 2 +- drivers/usb/host/xhci.c | 2 +- drivers/usb/isp1760/isp1760-core.c | 3 --- drivers/usb/isp1760/isp1760-if.c | 1 - drivers/usb/musb/musb_host.c | 2 +- drivers/usb/renesas_usbhs/mod_host.c | 2 +- include/linux/usb.h | 1 - include/linux/usb/hcd.h | 7 +++++-- 35 files changed, 31 insertions(+), 62 deletions(-)
drivers/staging/octeon-usb/octeon-hcd.c +1 -1
@@ -3512,7 +3512,7 @@ static const struct hc_driver octeon_hc_driver = { .product_desc = "Octeon Host Controller", .hcd_priv_size = sizeof(struct octeon_hcd), .irq = octeon_usb_irq, - .flags = HCD_MEMORY | HCD_USB2, + .flags = HCD_MEMORY | HCD_DMA | HCD_USB2, .start = octeon_usb_start, .stop = octeon_usb_stop, .urb_enqueue = octeon_usb_urb_enqueue,
drivers/usb/core/hcd.c +0 -1
@@ -2454,7 +2454,6 @@ struct usb_hcd *__usb_create_hcd(const struct hc_driver *driver, hcd->self.controller = dev; hcd->self.sysdev = sysdev; hcd->self.bus_name = bus_name; - hcd->self.uses_dma = (sysdev->dma_mask !=NULL );timer_setup (&hcd->rh_timer, rh_timer_func,0 );#ifdef CONFIG_PM
drivers/usb/dwc2/hcd.c +3 -3
@@ -5062,13 +5062,13 @@ int dwc2_hcd_init(struct dwc2_hsotg *hsotg) dwc2_hc_driver.reset_device = dwc2_reset_device; } +if (hsotg->params.host_dma) + dwc2_hc_driver.flags |=HCD_DMA ; + hcd =usb_create_hcd (&dwc2_hc_driver, hsotg->dev,dev_name (hsotg->dev));if (!hcd) goto error1; -if (!hsotg->params.host_dma) - hcd->self.uses_dma =0 ; - hcd->has_tt =1 ; res =platform_get_resource (pdev,IORESOURCE_MEM ,0 );
drivers/usb/host/ehci-grlib.c +1 -1
@@ -30,7 +30,7 @@ static const struct hc_driver ehci_grlib_hc_driver = {* generic hardware linkage */ .irq = ehci_irq, - .flags = HCD_MEMORY | HCD_USB2 | HCD_BH, + .flags = HCD_MEMORY | HCD_DMA | HCD_USB2 | HCD_BH,/* * basic lifecycle operations
drivers/usb/host/ehci-hcd.c +1 -1
@@ -1193,7 +1193,7 @@ static const struct hc_driver ehci_hc_driver = {* generic hardware linkage */ .irq = ehci_irq, - .flags = HCD_MEMORY | HCD_USB2 | HCD_BH, + .flags = HCD_MEMORY | HCD_DMA | HCD_USB2 | HCD_BH,/* * basic lifecycle operations
drivers/usb/host/ehci-pmcmsp.c +1 -1
@@ -250,7 +250,7 @@ static const struct hc_driver ehci_msp_hc_driver = {* generic hardware linkage */ .irq = ehci_irq, - .flags = HCD_MEMORY | HCD_USB2 | HCD_BH, + .flags = HCD_MEMORY | HCD_DMA | HCD_USB2 | HCD_BH,/* * basic lifecycle operations
drivers/usb/host/ehci-ppc-of.c +1 -1
@@ -31,7 +31,7 @@ static const struct hc_driver ehci_ppc_of_hc_driver = {* generic hardware linkage */ .irq = ehci_irq, - .flags = HCD_MEMORY | HCD_USB2 | HCD_BH, + .flags = HCD_MEMORY | HC_DMA | HCD_USB2 | HCD_BH,/* * basic lifecycle operations
drivers/usb/host/ehci-ps3.c +1 -1
@@ -59,7 +59,7 @@ static const struct hc_driver ps3_ehci_hc_driver = { .product_desc = "PS3 EHCI Host Controller", .hcd_priv_size = sizeof(struct ehci_hcd), .irq = ehci_irq, - .flags = HCD_MEMORY | HCD_USB2 | HCD_BH, + .flags = HCD_MEMORY | HCD_DMA | HCD_USB2 | HCD_BH, .reset = ps3_ehci_hc_reset, .start = ehci_run, .stop = ehci_stop,
drivers/usb/host/ehci-sh.c +1 -1
@@ -33,7 +33,7 @@ static const struct hc_driver ehci_sh_hc_driver = {* generic hardware linkage */ .irq = ehci_irq, - .flags = HCD_USB2 | HCD_MEMORY | HCD_BH, + .flags = HCD_USB2 | HCD_DMA | HCD_MEMORY | HCD_BH,/* * basic lifecycle operations
drivers/usb/host/ehci-xilinx-of.c +1 -1
@@ -66,7 +66,7 @@ static const struct hc_driver ehci_xilinx_of_hc_driver = {* generic hardware linkage */ .irq = ehci_irq, - .flags = HCD_MEMORY | HCD_USB2 | HCD_BH, + .flags = HCD_MEMORY | HCD_DMA | HCD_USB2 | HCD_BH,/* * basic lifecycle operations
drivers/usb/host/fhci-hcd.c +1 -1
@@ -538,7 +538,7 @@ static const struct hc_driver fhci_driver = {/* generic hardware linkage */ .irq = fhci_irq, - .flags = HCD_USB11 | HCD_MEMORY, + .flags = HCD_DMA | HCD_USB11 | HCD_MEMORY,/* basic lifecycle operation */ .start = fhci_start,
drivers/usb/host/fotg210-hcd.c +1 -1
@@ -5504,7 +5504,7 @@ static const struct hc_driver fotg210_fotg210_hc_driver = {* generic hardware linkage */ .irq = fotg210_irq, - .flags = HCD_MEMORY | HCD_USB2, + .flags = HCD_MEMORY | HCD_DMA | HCD_USB2,/* * basic lifecycle operations
drivers/usb/host/imx21-hcd.c +1 -1
@@ -1771,7 +1771,7 @@ static const struct hc_driver imx21_hc_driver = { .product_desc = "IMX21 USB Host Controller", .hcd_priv_size = sizeof(struct imx21), - .flags = HCD_USB11, + .flags = HCD_DMA | HCD_USB11, .irq = imx21_irq, .reset = imx21_hc_reset,
drivers/usb/host/isp116x-hcd.c +0 -6
@@ -1581,12 +1581,6 @@ static int isp116x_probe(struct platform_device *pdev) irq = ires->start; irqflags = ires->flags &IRQF_TRIGGER_MASK ; -if (pdev->dev.dma_mask) { -DBG ("DMA not supported\n" ); - ret = -EINVAL ; - goto err1; - } -if (!request_mem_region (addr->start,2 , hcd_name)) { ret = -EBUSY ; goto err1;
drivers/usb/host/isp1362-hcd.c +0 -5
@@ -2645,11 +2645,6 @@ static int isp1362_probe(struct platform_device *pdev)if (pdev->num_resources <3 )return -ENODEV ; -if (pdev->dev.dma_mask) { -DBG (1 ,"won't do DMA" ); -return -ENODEV ; - } - irq_res =platform_get_resource (pdev,IORESOURCE_IRQ ,0 );if (!irq_res)return -ENODEV ;
drivers/usb/host/ohci-hcd.c +1 -1
@@ -1178,7 +1178,7 @@ static const struct hc_driver ohci_hc_driver = {* generic hardware linkage */ .irq = ohci_irq, - .flags = HCD_MEMORY | HCD_USB11, + .flags = HCD_MEMORY | HCD_DMA | HCD_USB11,/* * basic lifecycle operations
drivers/usb/host/ohci-ppc-of.c +1 -1
@@ -50,7 +50,7 @@ static const struct hc_driver ohci_ppc_of_hc_driver = {* generic hardware linkage */ .irq = ohci_irq, - .flags = HCD_USB11 | HCD_MEMORY, + .flags = HCD_USB11 | HCD_DMA | HCD_MEMORY,/* * basic lifecycle operations
drivers/usb/host/ohci-ps3.c +1 -1
@@ -46,7 +46,7 @@ static const struct hc_driver ps3_ohci_hc_driver = { .product_desc = "PS3 OHCI Host Controller", .hcd_priv_size = sizeof(struct ohci_hcd), .irq = ohci_irq, - .flags = HCD_MEMORY | HCD_USB11, + .flags = HCD_MEMORY | HCD_DMA | HCD_USB11, .reset = ps3_ohci_hc_reset, .start = ps3_ohci_hc_start, .stop = ohci_stop,
drivers/usb/host/ohci-sa1111.c +1 -1
@@ -84,7 +84,7 @@ static const struct hc_driver ohci_sa1111_hc_driver = {* generic hardware linkage */ .irq = ohci_irq, - .flags = HCD_USB11 | HCD_MEMORY, + .flags = HCD_USB11 | HCD_DMA | HCD_MEMORY,/* * basic lifecycle operations
drivers/usb/host/ohci-sm501.c +1 -1
@@ -49,7 +49,7 @@ static const struct hc_driver ohci_sm501_hc_driver = {* generic hardware linkage */ .irq = ohci_irq, - .flags = HCD_USB11 | HCD_MEMORY, + .flags = HCD_USB11 | HCD_DMA | HCD_MEMORY,/* * basic lifecycle operations
drivers/usb/host/ohci-tmio.c +1 -1
@@ -153,7 +153,7 @@ static const struct hc_driver ohci_tmio_hc_driver = {/* generic hardware linkage */ .irq = ohci_irq, - .flags = HCD_USB11 | HCD_MEMORY, + .flags = HCD_USB11 | HCD_DMA | HCD_MEMORY,/* basic lifecycle operations */ .start = ohci_tmio_start,
drivers/usb/host/oxu210hp-hcd.c +0 -3
@@ -2649,9 +2649,6 @@ static int oxu_reset(struct usb_hcd *hcd)INIT_LIST_HEAD (&oxu->urb_list); oxu->urb_len =0 ; -/* FIMXE */ - hcd->self.controller->dma_mask =NULL ; -if (oxu->is_otg) { oxu->caps = hcd->regs +OXU_OTG_CAP_OFFSET ; oxu->regs = hcd->regs + OXU_OTG_CAP_OFFSET + \
drivers/usb/host/r8a66597-hcd.c +0 -6
@@ -2411,12 +2411,6 @@ static int r8a66597_probe(struct platform_device *pdev)if (usb_disabled ())return -ENODEV ; -if (pdev->dev.dma_mask) { - ret = -EINVAL ; -dev_err (&pdev->dev,"dma not supported\n" ); - goto clean_up; - } - res =platform_get_resource (pdev,IORESOURCE_MEM ,0 );if (!res) { ret = -ENODEV ;
drivers/usb/host/sl811-hcd.c +0 -6
@@ -1632,12 +1632,6 @@ sl811h_probe(struct platform_device *dev) irq = ires->start; irqflags = ires->flags &IRQF_TRIGGER_MASK ; -/* refuse to confuse usbcore */ -if (dev->dev.dma_mask) { -dev_dbg (&dev->dev,"no we won't dma\n" ); -return -EINVAL ; - } -/* the chip may be wired for either kind of addressing */ addr =platform_get_resource (dev,IORESOURCE_MEM ,0 ); data =platform_get_resource (dev,IORESOURCE_MEM ,1 );
drivers/usb/host/u132-hcd.c +0 -2
@@ -3077,8 +3077,6 @@ static int u132_probe(struct platform_device *pdev) retval =ftdi_read_pcimem (pdev, roothub.a, &rh_a);if (retval)return retval; -if (pdev->dev.dma_mask) -return -EINVAL ; hcd =usb_create_hcd (&u132_hc_driver, &pdev->dev,dev_name (&pdev->dev));if (!hcd) {
drivers/usb/host/uhci-grlib.c +1 -1
@@ -63,7 +63,7 @@ static const struct hc_driver uhci_grlib_hc_driver = {/* Generic hardware linkage */ .irq = uhci_irq, - .flags = HCD_MEMORY | HCD_USB11, + .flags = HCD_MEMORY | HCD_DMA | HCD_USB11,/* Basic lifecycle operations */ .reset = uhci_grlib_init,
drivers/usb/host/uhci-pci.c +1 -1
@@ -261,7 +261,7 @@ static const struct hc_driver uhci_driver = {/* Generic hardware linkage */ .irq = uhci_irq, - .flags = HCD_USB11, + .flags = HCD_DMA | HCD_USB11,/* Basic lifecycle operations */ .reset = uhci_pci_init,
drivers/usb/host/uhci-platform.c +1 -1
@@ -41,7 +41,7 @@ static const struct hc_driver uhci_platform_hc_driver = {/* Generic hardware linkage */ .irq = uhci_irq, - .flags = HCD_MEMORY | HCD_USB11, + .flags = HCD_MEMORY | HCD_DMA | HCD_USB11,/* Basic lifecycle operations */ .reset = uhci_platform_init,
drivers/usb/host/xhci.c +1 -1
@@ -5217,7 +5217,7 @@ static const struct hc_driver xhci_hc_driver = {* generic hardware linkage */ .irq = xhci_irq, - .flags = HCD_MEMORY | HCD_USB3 | HCD_SHARED, + .flags = HCD_MEMORY | HCD_DMA | HCD_USB3 | HCD_SHARED,/* * basic lifecycle operations
drivers/usb/isp1760/isp1760-core.c +0 -3
@@ -120,9 +120,6 @@ int isp1760_register(struct resource *mem, int irq, unsigned long irqflags, (!IS_ENABLED(CONFIG_USB_ISP1761_UDC) || udc_disabled))return -ENODEV ; -/* prevent usb-core allocating DMA pages */ - dev->dma_mask =NULL ; - isp =devm_kzalloc (dev,sizeof (*isp),GFP_KERNEL );if (!isp)return -ENOMEM ;
drivers/usb/isp1760/isp1760-if.c +0 -1
@@ -139,7 +139,6 @@ static int isp1761_pci_probe(struct pci_dev *dev,pci_set_master (dev); - dev->dev.dma_mask =NULL ; ret =isp1760_register (&dev->resource[3 ], dev->irq,0 , &dev->dev, devflags);if (ret <0 )
drivers/usb/musb/musb_host.c +1 -1
@@ -2689,7 +2689,7 @@ static const struct hc_driver musb_hc_driver = { .description = "musb-hcd", .product_desc = "MUSB HDRC host driver", .hcd_priv_size = sizeof(struct musb *), - .flags = HCD_USB2 | HCD_MEMORY, + .flags = HCD_USB2 | HCD_DMA | HCD_MEMORY,/* not using irq handler or reset hooks from usbcore, since * those must be shared with peripheral code for OTG configs
drivers/usb/renesas_usbhs/mod_host.c +1 -1
@@ -1283,7 +1283,7 @@ static const struct hc_driver usbhsh_driver = {/* * generic hardware linkage */ - .flags = HCD_USB2, + .flags = HCD_DMA | HCD_USB2, .start = usbhsh_host_start, .stop = usbhsh_host_stop,
include/linux/usb.h +0 -1
@@ -426,7 +426,6 @@ struct usb_bus {struct device *sysdev;/* as seen from firmware or bus */ int busnum;/* Bus number (in order of reg) */ const char *bus_name;/* stable id (PCI slot_name etc) */ -u8 uses_dma;/* Does the host controller use DMA? */ u8 uses_pio_for_control;/* * Does the host controller use PIO * for control transfers?
include/linux/usb/hcd.h +5 -2
@@ -256,6 +256,7 @@ struct hc_driver {int flags;#define HCD_MEMORY 0x0001/* HC regs use memory (else I/O) */ +#define HCD_DMA 0x0002/* HC uses DMA */ #define HCD_SHARED 0x0004/* Two (or more) usb_hcds share HW */ #define HCD_USB11 0x0010/* USB 1.1 */ #define HCD_USB2 0x0020/* USB 2.0 */
@@ -422,8 +423,10 @@ static inline bool hcd_periodic_completion_in_progress(struct usb_hcd *hcd,return hcd->high_prio_bh.completing_ep == ep; } -#define hcd_uses_dma (hcd) \ - (IS_ENABLED(CONFIG_HAS_DMA) && (hcd)->self.uses_dma) +static inline bool hcd_uses_dma (struct usb_hcd *hcd) +{ +return IS_ENABLED (CONFIG_HAS_DMA ) && (hcd->driver->flags &HCD_DMA ); +}extern int usb_hcd_link_urb_to_ep (struct usb_hcd *hcd,struct urb *urb); extern int usb_hcd_check_unlink_urb(struct usb_hcd *hcd, struct urb *urb,
Signature · 1 line
2.20.1
Re: [PATCH 3/6] usb: add a HCD_DMA flag instead of guestimating DMA capabilities
drivers/usb/host/ehci-ppc-of.c static const struct hc_driver ehci_ppc_of_hc_driver = { · line 34
Quoted 10 lines
drivers/usb/host/ehci-ppc-of.c +1
-1@@ -31,7 +31,7 @@ static const struct hc_driver ehci_ppc_of_hc_driver = {* generic hardware linkage */ .irq = ehci_irq, - .flags = HCD_MEMORY | HCD_USB2 | HCD_BH, + .flags = HCD_MEMORY | HC_DMA | HCD_USB2 | HCD_BH,
FYI, the kbuild bot found a little typo here, so even for the unlikely case that the series is otherwise perfect I'll have to resend it at least once.
PATCH 4/6 usb/max3421: remove the dummy {un,}map_urb_for_dma methods
In reply to Christoph HellwigNow that we have an explicit HCD_DMA flag, there is not need to override these methods.
--- drivers/usb/host/max3421-hcd.c | 17 ----------------- 1 file changed, 17 deletions(-)
drivers/usb/host/max3421-hcd.c +0 -17
@@ -1800,21 +1800,6 @@ max3421_bus_resume(struct usb_hcd *hcd)return -1 ; } -/* -* The SPI driver already takes care of DMA-mapping/unmapping, so no -* reason to do it twice. -*/ -static int -max3421_map_urb_for_dma (struct usb_hcd *hcd,struct urb *urb,gfp_t mem_flags) -{ -return 0 ; -} - -static void -max3421_unmap_urb_for_dma (struct usb_hcd *hcd,struct urb *urb) -{ -} -static const struct hc_driver max3421_hcd_desc = { .description ="max3421" , .product_desc =DRIVER_DESC ,
@@ -1826,8 +1811,6 @@ static const struct hc_driver max3421_hcd_desc = { .get_frame_number = max3421_get_frame_number, .urb_enqueue = max3421_urb_enqueue, .urb_dequeue = max3421_urb_dequeue, - .map_urb_for_dma = max3421_map_urb_for_dma, - .unmap_urb_for_dma = max3421_unmap_urb_for_dma, .endpoint_disable = max3421_endpoint_disable, .hub_status_data = max3421_hub_status_data, .hub_control = max3421_hub_control,
Signature · 1 line
2.20.1
PATCH 5/6 dma-mapping: remove is_device_dma_capable
In reply to Christoph HellwigNo users left.
--- include/linux/dma-mapping.h | 5 ----- 1 file changed, 5 deletions(-)
include/linux/dma-mapping.h +0 -5
@@ -149,11 +149,6 @@ static inline int valid_dma_direction(int dma_direction) (dma_direction == DMA_FROM_DEVICE)); } -static inline int is_device_dma_capable (struct device *dev) -{ -return dev->dma_mask !=NULL && *dev->dma_mask !=DMA_MASK_NONE ; -} -#ifdef CONFIG_DMA_DECLARE_COHERENT /* * These three functions are only for dma allocator.
Signature · 1 line
2.20.1
PATCH 6/6 driver core: initialize a default DMA mask for platform device
In reply to Christoph HellwigWe still treat devices without a DMA mask as defaulting to 32-bits for both mask, but a few releases ago we've started warning about such cases, as they require special cases to work around this sloppyness. Add a dma_mask field to struct platform_object so that we can initialize the dma_mask pointer in struct device and initialize both masks to 32-bits by default. Architectures can still override this in arch_setup_pdev_archdata if needed.
Note that the code looks a little odd with the various conditionals because we have to support platform_device structures that are statically allocated.
--- drivers/base/platform.c | 15 +++++++++++++-- include/linux/platform_device.h | 1 + 2 files changed, 14 insertions(+), 2 deletions(-)
drivers/base/platform.c +13 -2
@@ -264,6 +264,17 @@ struct platform_object {char name[]; }; +static void setup_pdev_archdata (struct platform_device *pdev) +{ +if (!pdev->dev.coherent_dma_mask) + pdev->dev.coherent_dma_mask =DMA_BIT_MASK (32 ); +if (!pdev->dma_mask) + pdev->dma_mask =DMA_BIT_MASK (32 ); +if (!pdev->dev.dma_mask) + pdev->dev.dma_mask = &pdev->dma_mask; +arch_setup_pdev_archdata (pdev); +}; +/** * platform_device_put - destroy a platform device * @pdev: platform device to free
@@ -310,7 +321,7 @@ struct platform_device *platform_device_alloc(const char *name, int id) pa->pdev.id = id;device_initialize (&pa->pdev.dev); pa->pdev.dev.release = platform_device_release; -arch_setup_pdev_archdata (&pa->pdev); +setup_pdev_archdata (&pa->pdev); }return pa ? &pa->pdev :NULL ;
@@ -512,7 +523,7 @@ EXPORT_SYMBOL_GPL(platform_device_del);int platform_device_register (struct platform_device *pdev) {device_initialize (&pdev->dev); -arch_setup_pdev_archdata (pdev); +setup_pdev_archdata (pdev);return platform_device_add (pdev); }EXPORT_SYMBOL_GPL (platform_device_register);
include/linux/platform_device.h +1 -0
@@ -24,6 +24,7 @@ struct platform_device {int id;bool id_auto;struct device dev; +u64 dma_mask;u32 num_resources;struct resource *resource;
Signature · 1 line
2.20.1
Robin Murphy
reviewerRe: [PATCH 6/6] driver core: initialize a default DMA mask for platform device
Quoted 4 lines · Christoph Hellwig
We still treat devices without a DMA mask as defaulting to 32-bits for both mask, but a few releases ago we've started warning about such cases, as they require special cases to work around this sloppyness. Add a
dma_maskfield to structplatform_objectso that we can initialize
s/object/device/
Quoted 7 lines
the
dma_maskpointer in struct device and initialize both masks to 32-bits by default. Architectures can still override this inarch_setup_pdev_archdataif needed.Note that the code looks a little odd with the various conditionals because we have to support
platform_devicestructures that are statically allocated.
This would be a good point to also get rid of the long-standing bodge in platform_device_register_full().
Quoted 11 lines
Signed-off-by Christoph Hellwig <[email protected]>--- drivers/base/platform.c | 15 +++++++++++++-- include/linux/platform_device.h | 1 + 2 files changed, 14 insertions(+), 2 deletions(-)drivers/base/platform.c +0
-0
drivers/base/platform.c struct platform_object { · line 267
Quoted 4 lines
drivers/base/platform.c +1
-0@@ -264,6 +264,17 @@ struct platform_object {char name[]; }; +static void setup_pdev_archdata (struct platform_device *pdev)
Bikeshed: painting the generic DMA API properties as "archdata" feels a bit off-target :/
drivers/base/platform.c struct platform_object { · line 275
Quoted 8 lines
drivers/base/platform.c +8
-0+{ +if (!pdev->dev.coherent_dma_mask) + pdev->dev.coherent_dma_mask =DMA_BIT_MASK (32 ); +if (!pdev->dma_mask) + pdev->dma_mask =DMA_BIT_MASK (32 ); +if (!pdev->dev.dma_mask) + pdev->dev.dma_mask = &pdev->dma_mask; +arch_setup_pdev_archdata (pdev);
AFAICS m68k's implementation of that arch hook becomes entirely redundant after this change, so may as well go. That would just leave powerpc's actual archdata, which at a glance looks like it could probably be cleaned up with not *too* much trouble.
Robin.
Quoted 36 lines
drivers/base/platform.c +4
-2+}; +/** * platform_device_put - destroy a platform device * @pdev: platform device to free @@ -310,7 +321,7 @@ struct platform_device *platform_device_alloc(const char *name, int id) pa->pdev.id = id;device_initialize (&pa->pdev.dev); pa->pdev.dev.release = platform_device_release; -arch_setup_pdev_archdata (&pa->pdev); +setup_pdev_archdata (&pa->pdev); }return pa ? &pa->pdev :NULL ;@@ -512,7 +523,7 @@ EXPORT_SYMBOL_GPL(platform_device_del);int platform_device_register (struct platform_device *pdev) {device_initialize (&pdev->dev); -arch_setup_pdev_archdata (pdev); +setup_pdev_archdata (pdev);return platform_device_add (pdev); }EXPORT_SYMBOL_GPL (platform_device_register);include/linux/platform_device.h +1
-0@@ -24,6 +24,7 @@ struct platform_device {int id;bool id_auto;struct device dev; +u64 dma_mask;u32 num_resources;struct resource *resource;
Re: [PATCH 6/6] driver core: initialize a default DMA mask for platform device
Quoted 5 lines · Robin Murphy
Quoted 2 lines
because we have to support
platform_devicestructures that are statically allocated.This would be a good point to also get rid of the long-standing bodge in
platform_device_register_full().
platform_device_register_full looks odd to start with, especially as the coumentation is rather lacking..
Quoted 5 lines
Quoted 1 line
+static void
setup_pdev_archdata(struct platform_device *pdev)Bikeshed: painting the generic DMA API properties as "archdata" feels a bit off-target :/
Quoted 13 lines
drivers/base/platform.c struct platform_object { · line 275
Quoted 8 lines
drivers/base/platform.c +8
-0+{ +if (!pdev->dev.coherent_dma_mask) + pdev->dev.coherent_dma_mask =DMA_BIT_MASK (32 ); +if (!pdev->dma_mask) + pdev->dma_mask =DMA_BIT_MASK (32 ); +if (!pdev->dev.dma_mask) + pdev->dev.dma_mask = &pdev->dma_mask; +arch_setup_pdev_archdata (pdev);AFAICS m68k's implementation of that arch hook becomes entirely redundant after this change, so may as well go. That would just leave powerpc's actual archdata, which at a glance looks like it could probably be cleaned up with not *too* much trouble.
Actually I think we can just kill both off. At the point archdata is indeed entirely misnamed.
Greg Kroah-Hartman
maintainerRe: [PATCH 6/6] driver core: initialize a default DMA mask for platform device
In reply to Christoph HellwigQuoted 63 lines · Christoph Hellwig
We still treat devices without a DMA mask as defaulting to 32-bits for both mask, but a few releases ago we've started warning about such cases, as they require special cases to work around this sloppyness. Add a
dma_maskfield to structplatform_objectso that we can initialize thedma_maskpointer in struct device and initialize both masks to 32-bits by default. Architectures can still override this inarch_setup_pdev_archdataif needed.Note that the code looks a little odd with the various conditionals because we have to support
platform_devicestructures that are statically allocated.Signed-off-by Christoph Hellwig <[email protected]>--- drivers/base/platform.c | 15 +++++++++++++-- include/linux/platform_device.h | 1 + 2 files changed, 14 insertions(+), 2 deletions(-)drivers/base/platform.c +13
-2@@ -264,6 +264,17 @@ struct platform_object {char name[]; }; +static void setup_pdev_archdata (struct platform_device *pdev) +{ +if (!pdev->dev.coherent_dma_mask) + pdev->dev.coherent_dma_mask =DMA_BIT_MASK (32 ); +if (!pdev->dma_mask) + pdev->dma_mask =DMA_BIT_MASK (32 ); +if (!pdev->dev.dma_mask) + pdev->dev.dma_mask = &pdev->dma_mask; +arch_setup_pdev_archdata (pdev); +}; +/** * platform_device_put - destroy a platform device * @pdev: platform device to free @@ -310,7 +321,7 @@ struct platform_device *platform_device_alloc(const char *name, int id) pa->pdev.id = id;device_initialize (&pa->pdev.dev); pa->pdev.dev.release = platform_device_release; -arch_setup_pdev_archdata (&pa->pdev); +setup_pdev_archdata (&pa->pdev); }return pa ? &pa->pdev :NULL ;@@ -512,7 +523,7 @@ EXPORT_SYMBOL_GPL(platform_device_del);int platform_device_register (struct platform_device *pdev) {device_initialize (&pdev->dev); -arch_setup_pdev_archdata (pdev); +setup_pdev_archdata (pdev);return platform_device_add (pdev); }EXPORT_SYMBOL_GPL (platform_device_register);include/linux/platform_device.h +0
-0
include/linux/platform_device.h struct platform_device { · line 27
Quoted 4 lines
include/linux/platform_device.h +1
-0@@ -24,6 +24,7 @@ struct platform_device {int id;bool id_auto;struct device dev; +u64 dma_mask;
Why is the dma_mask in 'struct device' which is part of this structure, not sufficient here? Shouldn't the "platform" be setting that up correctly already in the "archdata" type callback?
confused,
greg k-h
Re: [PATCH 6/6] driver core: initialize a default DMA mask for platform device
Quoted 11 lines · Greg Kroah-Hartman
include/linux/platform_device.h struct platform_device { · line 27
Quoted 7 lines
include/linux/platform_device.h +1
-0@@ -24,6 +24,7 @@ struct platform_device {int id;bool id_auto;struct device dev; +u64 dma_mask;Why is the
dma_maskin 'struct device' which is part of this structure, not sufficient here? Shouldn't the "platform" be setting that up correctly already in the "archdata" type callback?
Becaus the dma_mask in struct device is a pointer that needs to point to something, and this is the best space we can allocate for 'something'. m68k and powerpc currently do something roughly equivalent at the moment, while everyone else just has horrible, horrible hacks. As mentioned in the changelog the intent of this patch is that we treat platform devices like any other bus, where the bus allocates the space for the dma_mask. The long term plan is to eventually kill that weird pointer indirection that doesn't help anyone, but for that we need to sort out the basics first.
Greg Kroah-Hartman
maintainerRe: [PATCH 6/6] driver core: initialize a default DMA mask for platform device
Quoted 13 lines · Christoph Hellwig
Quoted 11 lines · Greg Kroah-Hartman
include/linux/platform_device.h struct platform_device { · line 27
Quoted 7 lines
include/linux/platform_device.h +1
-0@@ -24,6 +24,7 @@ struct platform_device {int id;bool id_auto;struct device dev; +u64 dma_mask;Why is the
dma_maskin 'struct device' which is part of this structure, not sufficient here? Shouldn't the "platform" be setting that up correctly already in the "archdata" type callback?
Quoted 9 lines
Becaus the
dma_maskin struct device is a pointer that needs to point to something, and this is the best space we can allocate for 'something'. m68k and powerpc currently do something roughly equivalent at the moment, while everyone else just has horrible, horrible hacks. As mentioned in the changelog the intent of this patch is that we treat platform devices like any other bus, where the bus allocates the space for thedma_mask. The long term plan is to eventually kill that weird pointer indirection that doesn't help anyone, but for that we need to sort out the basics first.
Ah, missed that, sorry. Ok, no objection from me. Might as well respin this series and I can queue it up after 5.3-rc5 is out (which will have your first 2 patches in it.)
thanks,
greg k-h
Greg Kroah-Hartman
maintainerQuoted 9 lines · Christoph Hellwig
Hi all,
this is another attempt to make sure the
dma_maskpointer is always initialized for platform devices. Not doing so lead to lots of boilerplate code, and makes platform devices different from all our major busses like PCI where we always set up adma_mask. In the long run this should also help to eventually makedma_maska scalar value instead of a pointer and remove even more cruft.
Quoted 8 lines
The bigger blocker for this last time was the fact that the usb subsystem uses the presence or lack of a
dma_maskto check if the core should do dma mapping for the driver, which is highly unusual. So we fix this first. Note that this has some overlap with the pending desire to use the properdma_mmap_coherenthelper for mapping usb buffers. The first two patches from this series should probably go into 5.3 and then uses as the basis for the decision to usedma_mmap_coherent.
I've taken the first 2 patches for 5.3-final. Given that patch 3 needs to be fixed, I'll wait for a respin of these before considering them.
thanks,
greg k-h
Quoted 2 lines · Greg Kroah-Hartman
I've taken the first 2 patches for 5.3-final. Given that patch 3 needs to be fixed, I'll wait for a respin of these before considering them.
I have a respun version ready, but I'd really like to hear some comments from usb developers about the approach before spamming everyone again..
Greg Kroah-Hartman
maintainerQuoted 7 lines · Christoph Hellwig
Quoted 2 lines · Greg Kroah-Hartman
I've taken the first 2 patches for 5.3-final. Given that patch 3 needs to be fixed, I'll wait for a respin of these before considering them.
I have a respun version ready, but I'd really like to hear some comments from usb developers about the approach before spamming everyone again..
Spam away, we can take it :)
Alan Stern
maintainerQuoted 7 lines · Christoph Hellwig
Quoted 2 lines · Greg Kroah-Hartman
I've taken the first 2 patches for 5.3-final. Given that patch 3 needs to be fixed, I'll wait for a respin of these before considering them.
I have a respun version ready, but I'd really like to hear some comments from usb developers about the approach before spamming everyone again..
I didn't see any problems with your approach at first glance; it looked like a good idea.
Alan Stern