| Message ID | 20250423-dt-memory-region-v2-v2-0-2fbd6ebd3c88@kernel.org |
|---|---|
| Headers | show |
| Series | of: Common "memory-region" parsing | expand |
Hello Rob, On 4/23/25 21:42, Rob Herring (Arm) wrote: > Drivers with "memory-region" properties currently have to do their own > parsing of "memory-region" properties. The result is all the drivers > have similar patterns of a call to parse "memory-region" and then get > the region's address and size. As this is a standard property, it should > have common functions for drivers to use. Add new functions to count the > number of regions and retrieve the region's address as a resource. > > Reviewed-by: Daniel Baluta <daniel.baluta@nxp.com> > Signed-off-by: Rob Herring (Arm) <robh@kernel.org> > --- > v2: > - Wrap function parameters > --- > drivers/of/of_reserved_mem.c | 80 +++++++++++++++++++++++++++++++++++++++++ > include/linux/of_reserved_mem.h | 26 ++++++++++++++ > 2 files changed, 106 insertions(+) > > diff --git a/drivers/of/of_reserved_mem.c b/drivers/of/of_reserved_mem.c > index ee2e31522d7e..f87915cce961 100644 > --- a/drivers/of/of_reserved_mem.c > +++ b/drivers/of/of_reserved_mem.c > @@ -12,6 +12,7 @@ > #define pr_fmt(fmt) "OF: reserved mem: " fmt > > #include <linux/err.h> > +#include <linux/ioport.h> > #include <linux/libfdt.h> > #include <linux/of.h> > #include <linux/of_fdt.h> > @@ -740,3 +741,82 @@ struct reserved_mem *of_reserved_mem_lookup(struct device_node *np) > return NULL; > } > EXPORT_SYMBOL_GPL(of_reserved_mem_lookup); > + > +/** > + * of_reserved_mem_region_to_resource() - Get a reserved memory region as a resource > + * @np: node containing 'memory-region' property > + * @idx: index of 'memory-region' property to lookup > + * @res: Pointer to a struct resource to fill in with reserved region > + * > + * This function allows drivers to lookup a node's 'memory-region' property > + * entries by index and return a struct resource for the entry. > + * > + * Returns 0 on success with @res filled in. Returns -ENODEV if 'memory-region' > + * is missing or unavailable, -EINVAL for any other error. > + */ > +int of_reserved_mem_region_to_resource(const struct device_node *np, > + unsigned int idx, struct resource *res) > +{ > + struct reserved_mem *rmem; > + > + if (!np) > + return -EINVAL; > + > + struct device_node __free(device_node) *target = of_parse_phandle(np, "memory-region", idx); > + if (!target || !of_device_is_available(target)) > + return -ENODEV; > + > + rmem = of_reserved_mem_lookup(target); > + if (!rmem) > + return -EINVAL; > + > + resource_set_range(res, rmem->base, rmem->size); > + res->name = rmem->name; > + return 0; > +} > +EXPORT_SYMBOL_GPL(of_reserved_mem_region_to_resource); > + > +/** > + * of_reserved_mem_region_to_resource_byname() - Get a reserved memory region as a resource > + * @np: node containing 'memory-region' property > + * @name: name of 'memory-region' property entry to lookup > + * @res: Pointer to a struct resource to fill in with reserved region > + * > + * This function allows drivers to lookup a node's 'memory-region' property > + * entries by name and return a struct resource for the entry. > + * > + * Returns 0 on success with @res filled in. Returns -ENODEV if 'memory-region' > + * is missing or unavailable, -EINVAL for any other error. of_property_match_string() can return some other error values. Thanks, Arnaud > + */ > +int of_reserved_mem_region_to_resource_byname(const struct device_node *np, > + const char *name, > + struct resource *res) > +{ > + int idx; > + > + if (!name) > + return -EINVAL; > + > + idx = of_property_match_string(np, "memory-region-names", name); > + if (idx < 0) > + return idx; > + > + return of_reserved_mem_region_to_resource(np, idx, res); > +} > +EXPORT_SYMBOL_GPL(of_reserved_mem_region_to_resource_byname); > + > +/** > + * of_reserved_mem_region_count() - Return the number of 'memory-region' entries > + * @np: node containing 'memory-region' property > + * > + * This function allows drivers to retrieve the number of entries for a node's > + * 'memory-region' property. > + * > + * Returns the number of entries on success, or negative error code on a > + * malformed property. > + */ > +int of_reserved_mem_region_count(const struct device_node *np) > +{ > + return of_count_phandle_with_args(np, "memory-region", NULL); > +} > +EXPORT_SYMBOL_GPL(of_reserved_mem_region_count); > diff --git a/include/linux/of_reserved_mem.h b/include/linux/of_reserved_mem.h > index e338282da652..f573423359f4 100644 > --- a/include/linux/of_reserved_mem.h > +++ b/include/linux/of_reserved_mem.h > @@ -7,6 +7,7 @@ > > struct of_phandle_args; > struct reserved_mem_ops; > +struct resource; > > struct reserved_mem { > const char *name; > @@ -39,6 +40,12 @@ int of_reserved_mem_device_init_by_name(struct device *dev, > void of_reserved_mem_device_release(struct device *dev); > > struct reserved_mem *of_reserved_mem_lookup(struct device_node *np); > +int of_reserved_mem_region_to_resource(const struct device_node *np, > + unsigned int idx, struct resource *res); > +int of_reserved_mem_region_to_resource_byname(const struct device_node *np, > + const char *name, struct resource *res); > +int of_reserved_mem_region_count(const struct device_node *np); > + > #else > > #define RESERVEDMEM_OF_DECLARE(name, compat, init) \ > @@ -63,6 +70,25 @@ static inline struct reserved_mem *of_reserved_mem_lookup(struct device_node *np > { > return NULL; > } > + > +static inline int of_reserved_mem_region_to_resource(const struct device_node *np, > + unsigned int idx, > + struct resource *res) > +{ > + return -ENOSYS; > +} > + > +static inline int of_reserved_mem_region_to_resource_byname(const struct device_node *np, > + const char *name, > + struct resource *res) > +{ > + return -ENOSYS; > +} > + > +static inline int of_reserved_mem_region_count(const struct device_node *np) > +{ > + return 0; > +} > #endif > > /** >
On 4/23/25 21:42, Rob Herring (Arm) wrote: > While there's a common function to parse "memory-region" properties for > DMA pool regions, there's not anything for driver private regions. As a > result, drivers have resorted to parsing "memory-region" properties > themselves repeating the same pattern over and over. To fix this, this > series adds 2 functions to handle those cases: > of_reserved_mem_region_to_resource() and of_reserved_mem_region_count(). > > I've converted the whole tree, but just including remoteproc here as > it has the most cases. I intend to apply the first 3 patches for 6.16 > so the driver conversions can be applied for 6.17. > > A git tree with all the drivers converted is here[1]. > > v2: > - Fix of_dma_set_restricted_buffer() to maintain behavior on warning msg > - Export devm_ioremap_resource_wc() > - Rework handling of resource name to drop unit-address from name as it > was before. > - Link to v1: > https://lore.kernel.org/all/20250317232426.952188-1-robh@kernel.org > > [1] git://git.kernel.org/pub/scm/linux/kernel/git/robh/linux.git dt/memory-region > > Signed-off-by: Rob Herring (Arm) <robh@kernel.org> > --- > Rob Herring (Arm) (4): > of: reserved_mem: Add functions to parse "memory-region" > of: Simplify of_dma_set_restricted_buffer() to use of_for_each_phandle() > devres: Export devm_ioremap_resource_wc() > remoteproc: Use of_reserved_mem_region_* functions for "memory-region" > > drivers/of/device.c | 31 +++++------- > drivers/of/of_reserved_mem.c | 80 +++++++++++++++++++++++++++++++ > drivers/remoteproc/imx_dsp_rproc.c | 45 +++++++---------- > drivers/remoteproc/imx_rproc.c | 68 +++++++++++--------------- > drivers/remoteproc/qcom_q6v5_adsp.c | 24 ++++------ > drivers/remoteproc/qcom_q6v5_mss.c | 60 ++++++++--------------- > drivers/remoteproc/qcom_q6v5_pas.c | 69 ++++++++++---------------- > drivers/remoteproc/qcom_q6v5_wcss.c | 25 ++++------ > drivers/remoteproc/qcom_wcnss.c | 23 ++++----- > drivers/remoteproc/rcar_rproc.c | 36 ++++++-------- > drivers/remoteproc/st_remoteproc.c | 41 ++++++++-------- > drivers/remoteproc/stm32_rproc.c | 44 ++++++++--------- > drivers/remoteproc/ti_k3_dsp_remoteproc.c | 28 +++++------ > drivers/remoteproc/ti_k3_m4_remoteproc.c | 28 +++++------ > drivers/remoteproc/ti_k3_r5_remoteproc.c | 28 +++++------ > drivers/remoteproc/xlnx_r5_remoteproc.c | 51 ++++++++------------ > include/linux/of_reserved_mem.h | 26 ++++++++++ > lib/devres.c | 1 + > 18 files changed, 339 insertions(+), 369 deletions(-) > --- > base-commit: 0af2f6be1b4281385b618cb86ad946eded089ac8 > change-id: 20250423-dt-memory-region-v2-a2b15caacc63 > Testing of the series on the STM32MP15 platform has passed. just one minor comment in patch 1/4. Acked-by: Arnaud Pouliquen <arnaud.pouliquen@foss.st.com> Thanks, Arnaud > Best regards,
On Fri, Apr 25, 2025 at 6:34 PM Iuliana Prodan <iuliana.prodan@nxp.com> wrote: > > Hello Mathieu, Rob, > > I've tested imx_dsp_rproc and it fails with: > > [ 39.743770] Unable to handle kernel paging request at virtual address > ffffffffffffffea > ... > [ 39.805078] Hardware name: NXP i.MX8MPlus EVK board (DT) > [ 39.810390] pstate: 00000005 (nzcv daif -PAN -UAO -TCO -DIT -SSBS > BTYPE=--) > [ 39.817353] pc : __pi_memset_generic+0x50/0x188 > [ 39.821892] lr : imx_dsp_rproc_prepare+0x3a4/0xea4 [imx_dsp_rproc] > [ 39.828079] sp : ffff8000853dbb10 > [ 39.831396] x29: ffff8000853dbb90 x28: 0000000092400000 x27: > ffff80007a36d3d8 > [ 39.838540] x26: ffff0000d0a5e410 x25: ffff80007a36d3f0 x24: > 0000000000000004 > [ 39.845685] x23: ffff0000d5414880 x22: ffff0000d5e1dce8 x21: > 0000000000000000 > [ 39.852827] x20: ffff0000d5e1db98 x19: ffff0000d5414b80 x18: > 00000000ffffffff > [ 39.859970] x17: 202c656c69687720 x16: 3e2074756f657672 x15: > ffff800081f8d050 > [ 39.867114] x14: ffff0000db584680 x13: 0000000000000003 x12: > 00007fffa3330000 > [ 39.874257] x11: 0000000000000004 x10: 0000000000000ab0 x9 : > 0000000000000000 > [ 39.881400] x8 : ffffffffffffffea x7 : 0000000000000000 x6 : > 000000000000003f > [ 39.888546] x5 : 0000000000000040 x4 : 0000000000000006 x3 : > 0000000000000004 > [ 39.895689] x2 : 0000000000008000 x1 : 0000000000000000 x0 : > ffffffffffffffea > [ 39.902837] Call trace: > [ 39.905284] __pi_memset_generic+0x50/0x188 (P) > [ 39.909821] rproc_boot+0x2c0/0x524 > [ 39.913317] state_store+0x40/0x100 > [ 39.916812] dev_attr_store+0x18/0x2c > [ 39.920478] sysfs_kf_write+0x7c/0x94 > [ 39.924146] kernfs_fop_write_iter+0x120/0x1e8 > [ 39.928598] vfs_write+0x244/0x37c > [ 39.932008] ksys_write+0x70/0x110 > [ 39.935413] __arm64_sys_write+0x1c/0x28 > [ 39.939342] invoke_syscall+0x48/0x104 > [ 39.943094] el0_svc_common.constprop.0+0xc0/0xe0 > [ 39.947805] do_el0_svc+0x1c/0x28 > [ 39.951123] el0_svc+0x30/0xcc > [ 39.954188] el0t_64_sync_handler+0x10c/0x138 > [ 39.958549] el0t_64_sync+0x198/0x19c > [ 39.962222] Code: d65f03c0 cb0803e4 f2400c84 54000080 (a9001d07) > [ 39.968317] ---[ end trace 0000000000000000 ]--- > > The problem seems to be when computing `cpu_addr = > devm_ioremap_resource_wc(dev, &res);`, in patch 4. > In `__devm_ioremap_resource` (see [1]), it's expecting the resource type > to be `IORESOURCE_MEM`, which is not the case here (at least the flags > are nowhere set for this). Thanks for testing and diagnosing. IORESOURCE_MEM needs to be set. Can you test with this change: index f87915cce961..00c93585758f 100644 --- a/drivers/of/of_reserved_mem.c +++ b/drivers/of/of_reserved_mem.c @@ -771,6 +771,7 @@ int of_reserved_mem_region_to_resource(const struct device_node *np, return -EINVAL; resource_set_range(res, rmem->base, rmem->size); + res->flags = IORESOURCE_MEM; res->name = rmem->name; return 0; }
On 4/28/2025 3:23 PM, Rob Herring wrote: > On Fri, Apr 25, 2025 at 6:34 PM Iuliana Prodan <iuliana.prodan@nxp.com> wrote: >> Hello Mathieu, Rob, >> >> I've tested imx_dsp_rproc and it fails with: >> >> [ 39.743770] Unable to handle kernel paging request at virtual address >> ffffffffffffffea >> ... >> [ 39.805078] Hardware name: NXP i.MX8MPlus EVK board (DT) >> [ 39.810390] pstate: 00000005 (nzcv daif -PAN -UAO -TCO -DIT -SSBS >> BTYPE=--) >> [ 39.817353] pc : __pi_memset_generic+0x50/0x188 >> [ 39.821892] lr : imx_dsp_rproc_prepare+0x3a4/0xea4 [imx_dsp_rproc] >> [ 39.828079] sp : ffff8000853dbb10 >> [ 39.831396] x29: ffff8000853dbb90 x28: 0000000092400000 x27: >> ffff80007a36d3d8 >> [ 39.838540] x26: ffff0000d0a5e410 x25: ffff80007a36d3f0 x24: >> 0000000000000004 >> [ 39.845685] x23: ffff0000d5414880 x22: ffff0000d5e1dce8 x21: >> 0000000000000000 >> [ 39.852827] x20: ffff0000d5e1db98 x19: ffff0000d5414b80 x18: >> 00000000ffffffff >> [ 39.859970] x17: 202c656c69687720 x16: 3e2074756f657672 x15: >> ffff800081f8d050 >> [ 39.867114] x14: ffff0000db584680 x13: 0000000000000003 x12: >> 00007fffa3330000 >> [ 39.874257] x11: 0000000000000004 x10: 0000000000000ab0 x9 : >> 0000000000000000 >> [ 39.881400] x8 : ffffffffffffffea x7 : 0000000000000000 x6 : >> 000000000000003f >> [ 39.888546] x5 : 0000000000000040 x4 : 0000000000000006 x3 : >> 0000000000000004 >> [ 39.895689] x2 : 0000000000008000 x1 : 0000000000000000 x0 : >> ffffffffffffffea >> [ 39.902837] Call trace: >> [ 39.905284] __pi_memset_generic+0x50/0x188 (P) >> [ 39.909821] rproc_boot+0x2c0/0x524 >> [ 39.913317] state_store+0x40/0x100 >> [ 39.916812] dev_attr_store+0x18/0x2c >> [ 39.920478] sysfs_kf_write+0x7c/0x94 >> [ 39.924146] kernfs_fop_write_iter+0x120/0x1e8 >> [ 39.928598] vfs_write+0x244/0x37c >> [ 39.932008] ksys_write+0x70/0x110 >> [ 39.935413] __arm64_sys_write+0x1c/0x28 >> [ 39.939342] invoke_syscall+0x48/0x104 >> [ 39.943094] el0_svc_common.constprop.0+0xc0/0xe0 >> [ 39.947805] do_el0_svc+0x1c/0x28 >> [ 39.951123] el0_svc+0x30/0xcc >> [ 39.954188] el0t_64_sync_handler+0x10c/0x138 >> [ 39.958549] el0t_64_sync+0x198/0x19c >> [ 39.962222] Code: d65f03c0 cb0803e4 f2400c84 54000080 (a9001d07) >> [ 39.968317] ---[ end trace 0000000000000000 ]--- >> >> The problem seems to be when computing `cpu_addr = >> devm_ioremap_resource_wc(dev, &res);`, in patch 4. >> In `__devm_ioremap_resource` (see [1]), it's expecting the resource type >> to be `IORESOURCE_MEM`, which is not the case here (at least the flags >> are nowhere set for this). > Thanks for testing and diagnosing. IORESOURCE_MEM needs to be set. Can > you test with this change: > > index f87915cce961..00c93585758f 100644 > --- a/drivers/of/of_reserved_mem.c > +++ b/drivers/of/of_reserved_mem.c > @@ -771,6 +771,7 @@ int of_reserved_mem_region_to_resource(const > struct device_node *np, > return -EINVAL; > > resource_set_range(res, rmem->base, rmem->size); > + res->flags = IORESOURCE_MEM; > res->name = rmem->name; > return 0; > } I've tested with this change, and is all good now. Iulia
While there's a common function to parse "memory-region" properties for DMA pool regions, there's not anything for driver private regions. As a result, drivers have resorted to parsing "memory-region" properties themselves repeating the same pattern over and over. To fix this, this series adds 2 functions to handle those cases: of_reserved_mem_region_to_resource() and of_reserved_mem_region_count(). I've converted the whole tree, but just including remoteproc here as it has the most cases. I intend to apply the first 3 patches for 6.16 so the driver conversions can be applied for 6.17. A git tree with all the drivers converted is here[1]. v2: - Fix of_dma_set_restricted_buffer() to maintain behavior on warning msg - Export devm_ioremap_resource_wc() - Rework handling of resource name to drop unit-address from name as it was before. - Link to v1: https://lore.kernel.org/all/20250317232426.952188-1-robh@kernel.org [1] git://git.kernel.org/pub/scm/linux/kernel/git/robh/linux.git dt/memory-region Signed-off-by: Rob Herring (Arm) <robh@kernel.org> --- Rob Herring (Arm) (4): of: reserved_mem: Add functions to parse "memory-region" of: Simplify of_dma_set_restricted_buffer() to use of_for_each_phandle() devres: Export devm_ioremap_resource_wc() remoteproc: Use of_reserved_mem_region_* functions for "memory-region" drivers/of/device.c | 31 +++++------- drivers/of/of_reserved_mem.c | 80 +++++++++++++++++++++++++++++++ drivers/remoteproc/imx_dsp_rproc.c | 45 +++++++---------- drivers/remoteproc/imx_rproc.c | 68 +++++++++++--------------- drivers/remoteproc/qcom_q6v5_adsp.c | 24 ++++------ drivers/remoteproc/qcom_q6v5_mss.c | 60 ++++++++--------------- drivers/remoteproc/qcom_q6v5_pas.c | 69 ++++++++++---------------- drivers/remoteproc/qcom_q6v5_wcss.c | 25 ++++------ drivers/remoteproc/qcom_wcnss.c | 23 ++++----- drivers/remoteproc/rcar_rproc.c | 36 ++++++-------- drivers/remoteproc/st_remoteproc.c | 41 ++++++++-------- drivers/remoteproc/stm32_rproc.c | 44 ++++++++--------- drivers/remoteproc/ti_k3_dsp_remoteproc.c | 28 +++++------ drivers/remoteproc/ti_k3_m4_remoteproc.c | 28 +++++------ drivers/remoteproc/ti_k3_r5_remoteproc.c | 28 +++++------ drivers/remoteproc/xlnx_r5_remoteproc.c | 51 ++++++++------------ include/linux/of_reserved_mem.h | 26 ++++++++++ lib/devres.c | 1 + 18 files changed, 339 insertions(+), 369 deletions(-) --- base-commit: 0af2f6be1b4281385b618cb86ad946eded089ac8 change-id: 20250423-dt-memory-region-v2-a2b15caacc63 Best regards,