| Message ID | 20190815233618.31630-2-sstabellini@kernel.org (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | [v6,1/8] xen/arm: pass node to device_tree_for_each_node | expand |
Hi, On 16/08/2019 00:36, Stefano Stabellini wrote: > Change the signature of process_memory_node to match > device_tree_node_func. Thanks to this change, the next patch will be > able to use device_tree_for_each_node to call process_memory_node on all > the children of a provided node. > > Return error if there is no reg property or if nr_banks is reached. Let > the caller deal with the error. This sentence does not match the change below. Only 2 of the new error paths are described here. > > Add a printk when device tree parsing fails. > > Signed-off-by: Stefano Stabellini <stefanos@xilinx.com> > --- > Changes in v6: > - fix out of space check > - bring back printk when address_cells or size_cells are not properly set > - return -EINVAL in that case (different from reg missing) > - add printk when parsing fails > - return -ENOENT when memory size is 0 > > Changes in v5: > - return -ENOENT if address_cells or size_cells are not properly set > > Changes in v4: > - return error if there is no reg propery, remove printk > - return error if nr_banks is reached > > Changes in v3: > - improve commit message > - check return value of process_memory_node > > Changes in v2: > - new > --- > xen/arch/arm/bootfdt.c | 29 ++++++++++++++++++----------- > 1 file changed, 18 insertions(+), 11 deletions(-) > > diff --git a/xen/arch/arm/bootfdt.c b/xen/arch/arm/bootfdt.c > index f1614ef7fc..9dc2c1352d 100644 > --- a/xen/arch/arm/bootfdt.c > +++ b/xen/arch/arm/bootfdt.c > @@ -130,9 +130,10 @@ int __init device_tree_for_each_node(const void *fdt, int node, > return 0; > } > > -static void __init process_memory_node(const void *fdt, int node, > - const char *name, > - u32 address_cells, u32 size_cells) > +static int __init process_memory_node(const void *fdt, int node, > + const char *name, int depth, > + u32 address_cells, u32 size_cells, > + void *data) > { > const struct fdt_property *prop; > int i; > @@ -145,15 +146,12 @@ static void __init process_memory_node(const void *fdt, int node, > { > printk("fdt: node `%s': invalid #address-cells or #size-cells", > name); > - return; > + return -EINVAL; > } > > prop = fdt_get_property(fdt, node, "reg", NULL); > if ( !prop ) > - { > - printk("fdt: node `%s': missing `reg' property\n", name); > - return; > - } > + return -ENOENT; > > cell = (const __be32 *)prop->data; > banks = fdt32_to_cpu(prop->len) / (reg_cells * sizeof (u32)); > @@ -162,11 +160,15 @@ static void __init process_memory_node(const void *fdt, int node, > { > device_tree_get_reg(&cell, address_cells, size_cells, &start, &size); > if ( !size ) > - continue; > + return -ENOENT; I don't think we can treat the same way the lack of "regs" properties and a size of 0. The former is expected as binding allow you to do it for reserved-memory. The latter is the user not writing the property correctly. So ignoring the latter will result to Xen potentially missing some reserved-regions (not great!). So, similar to #address-cells/#size-cells discussion, we should return an error we are able to distinguish. Probably -EINVAL. Cheers,
On Fri, 16 Aug 2019, Julien Grall wrote: > Hi, > > On 16/08/2019 00:36, Stefano Stabellini wrote: > > Change the signature of process_memory_node to match > > device_tree_node_func. Thanks to this change, the next patch will be > > able to use device_tree_for_each_node to call process_memory_node on all > > the children of a provided node. > > > > Return error if there is no reg property or if nr_banks is reached. Let > > the caller deal with the error. > > This sentence does not match the change below. Only 2 of the new error paths > are described here. > > > > > Add a printk when device tree parsing fails. > > > > Signed-off-by: Stefano Stabellini <stefanos@xilinx.com> > > --- > > Changes in v6: > > - fix out of space check > > - bring back printk when address_cells or size_cells are not properly set > > - return -EINVAL in that case (different from reg missing) > > - add printk when parsing fails > > - return -ENOENT when memory size is 0 > > > > Changes in v5: > > - return -ENOENT if address_cells or size_cells are not properly set > > > > Changes in v4: > > - return error if there is no reg propery, remove printk > > - return error if nr_banks is reached > > > > Changes in v3: > > - improve commit message > > - check return value of process_memory_node > > > > Changes in v2: > > - new > > --- > > xen/arch/arm/bootfdt.c | 29 ++++++++++++++++++----------- > > 1 file changed, 18 insertions(+), 11 deletions(-) > > > > diff --git a/xen/arch/arm/bootfdt.c b/xen/arch/arm/bootfdt.c > > index f1614ef7fc..9dc2c1352d 100644 > > --- a/xen/arch/arm/bootfdt.c > > +++ b/xen/arch/arm/bootfdt.c > > @@ -130,9 +130,10 @@ int __init device_tree_for_each_node(const void *fdt, > > int node, > > return 0; > > } > > -static void __init process_memory_node(const void *fdt, int node, > > - const char *name, > > - u32 address_cells, u32 size_cells) > > +static int __init process_memory_node(const void *fdt, int node, > > + const char *name, int depth, > > + u32 address_cells, u32 size_cells, > > + void *data) > > { > > const struct fdt_property *prop; > > int i; > > @@ -145,15 +146,12 @@ static void __init process_memory_node(const void > > *fdt, int node, > > { > > printk("fdt: node `%s': invalid #address-cells or #size-cells", > > name); > > - return; > > + return -EINVAL; > > } > > prop = fdt_get_property(fdt, node, "reg", NULL); > > if ( !prop ) > > - { > > - printk("fdt: node `%s': missing `reg' property\n", name); > > - return; > > - } > > + return -ENOENT; > > cell = (const __be32 *)prop->data; > > banks = fdt32_to_cpu(prop->len) / (reg_cells * sizeof (u32)); > > @@ -162,11 +160,15 @@ static void __init process_memory_node(const void > > *fdt, int node, > > { > > device_tree_get_reg(&cell, address_cells, size_cells, &start, > > &size); > > if ( !size ) > > - continue; > > + return -ENOENT; > > I don't think we can treat the same way the lack of "regs" properties and a > size of 0. > > The former is expected as binding allow you to do it for reserved-memory. The > latter is the user not writing the property correctly. So ignoring the latter > will result to Xen potentially missing some reserved-regions (not great!). > > So, similar to #address-cells/#size-cells discussion, we should return an > error we are able to distinguish. Probably -EINVAL. I think you are right, and honestly I was thinking about it while I updated this patch. If I use -EINVAL, it would be the same return error as the "invalid #address-cells or #size-cells". I just wanted to double-check that it is OK for you.
On 8/17/19 1:48 AM, Stefano Stabellini wrote: > On Fri, 16 Aug 2019, Julien Grall wrote: > I think you are right, and honestly I was thinking about it while I > updated this patch. If I use -EINVAL, it would be the same return error > as the "invalid #address-cells or #size-cells". I just wanted to > double-check that it is OK for you. We don't need to differentiate "invalid #{address, size}-cells" from "zero size", so using the same errno is fine. Cheers,
diff --git a/xen/arch/arm/bootfdt.c b/xen/arch/arm/bootfdt.c index f1614ef7fc..9dc2c1352d 100644 --- a/xen/arch/arm/bootfdt.c +++ b/xen/arch/arm/bootfdt.c @@ -130,9 +130,10 @@ int __init device_tree_for_each_node(const void *fdt, int node, return 0; } -static void __init process_memory_node(const void *fdt, int node, - const char *name, - u32 address_cells, u32 size_cells) +static int __init process_memory_node(const void *fdt, int node, + const char *name, int depth, + u32 address_cells, u32 size_cells, + void *data) { const struct fdt_property *prop; int i; @@ -145,15 +146,12 @@ static void __init process_memory_node(const void *fdt, int node, { printk("fdt: node `%s': invalid #address-cells or #size-cells", name); - return; + return -EINVAL; } prop = fdt_get_property(fdt, node, "reg", NULL); if ( !prop ) - { - printk("fdt: node `%s': missing `reg' property\n", name); - return; - } + return -ENOENT; cell = (const __be32 *)prop->data; banks = fdt32_to_cpu(prop->len) / (reg_cells * sizeof (u32)); @@ -162,11 +160,15 @@ static void __init process_memory_node(const void *fdt, int node, { device_tree_get_reg(&cell, address_cells, size_cells, &start, &size); if ( !size ) - continue; + return -ENOENT; bootinfo.mem.bank[bootinfo.mem.nr_banks].start = start; bootinfo.mem.bank[bootinfo.mem.nr_banks].size = size; bootinfo.mem.nr_banks++; } + + if ( i < banks ) + return -ENOSPC; + return 0; } static void __init process_multiboot_node(const void *fdt, int node, @@ -298,15 +300,20 @@ static int __init early_scan_node(const void *fdt, u32 address_cells, u32 size_cells, void *data) { + int rc = 0; + if ( device_tree_node_matches(fdt, node, "memory") ) - process_memory_node(fdt, node, name, address_cells, size_cells); + rc = process_memory_node(fdt, node, name, depth, + address_cells, size_cells, NULL); else if ( depth <= 3 && (device_tree_node_compatible(fdt, node, "xen,multiboot-module" ) || device_tree_node_compatible(fdt, node, "multiboot,module" ))) process_multiboot_node(fdt, node, name, address_cells, size_cells); else if ( depth == 1 && device_tree_node_matches(fdt, node, "chosen") ) process_chosen_node(fdt, node, name, address_cells, size_cells); - return 0; + if ( rc < 0 ) + printk("fdt: node `%s': parsing failed\n", name); + return rc; } static void __init early_print_info(void)
Change the signature of process_memory_node to match device_tree_node_func. Thanks to this change, the next patch will be able to use device_tree_for_each_node to call process_memory_node on all the children of a provided node. Return error if there is no reg property or if nr_banks is reached. Let the caller deal with the error. Add a printk when device tree parsing fails. Signed-off-by: Stefano Stabellini <stefanos@xilinx.com> --- Changes in v6: - fix out of space check - bring back printk when address_cells or size_cells are not properly set - return -EINVAL in that case (different from reg missing) - add printk when parsing fails - return -ENOENT when memory size is 0 Changes in v5: - return -ENOENT if address_cells or size_cells are not properly set Changes in v4: - return error if there is no reg propery, remove printk - return error if nr_banks is reached Changes in v3: - improve commit message - check return value of process_memory_node Changes in v2: - new --- xen/arch/arm/bootfdt.c | 29 ++++++++++++++++++----------- 1 file changed, 18 insertions(+), 11 deletions(-)