Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1237364 > unrolled thread

[PATCH v2] of/fdt: Allow memory node and root to have different size/address cells

Back to article view | Back to linux.kernel

Contents

  [PATCH v2] of/fdt: Allow memory node and root to have different size/address cells Vineet Gupta <Vineet.Gupta1@synopsys.com> - 2015-10-01 14:20 +0200
    Re: [PATCH v2] of/fdt: Allow memory node and root to have different size/address cells Arnd Bergmann <arnd@arndb.de> - 2015-10-01 14:30 +0200
      Re: [PATCH v2] of/fdt: Allow memory node and root to have different  size/address cells Vineet Gupta <Vineet.Gupta1@synopsys.com> - 2015-10-03 10:10 +0200

#1237364 — [PATCH v2] of/fdt: Allow memory node and root to have different size/address cells

Currently memory node parsing uses root "#size-cells", "#address-cells"
This doesn't work correctly when memory address/size is different or
greater than root's.

e.g. ARC 32-bit systems implementing physical adressing extension and
say 4GB of memory. All peripherals mappings stay within the 4GB (so root
address/size cells remain 1 each), only the memory node address/size
cells needs to specify greater than 32-bits as below

	memory {
		device_type = "memory";
		reg = <0x0 0x80000000 0x1 0x00000000>;	/* 4 GB */
		#address-cells = <2>;
		#size-cells = <2>;
	};

This patch lets me boot a ARC system with PAE40 + 4GB of memory specified
as above and fails to boot otherwise as memory parsing doesn't populate
right base, size.

---
Changes since v2:
   silly typo creeped in due to last minute change to use root props
---
Cc: Grant Likely <grant.likely@linaro.org>
Cc: Rob Herring <robh+dt@kernel.org>
Cc: devicetree@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Cc: linux-arch@vger.kernel.org
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
---
 drivers/of/fdt.c | 21 ++++++++++++++++-----
 1 file changed, 16 insertions(+), 5 deletions(-)

diff --git a/drivers/of/fdt.c b/drivers/of/fdt.c
index 07496560e5b9..48618bb2674a 100644
--- a/drivers/of/fdt.c
+++ b/drivers/of/fdt.c
@@ -888,8 +888,8 @@ int __init early_init_dt_scan_memory(unsigned long node, const char *uname,
 				     int depth, void *data)
 {
 	const char *type = of_get_flat_dt_prop(node, "device_type", NULL);
-	const __be32 *reg, *endp;
-	int l;
+	const __be32 *reg, *endp, *prop;
+	int l, dt_mem_addr_cells, dt_mem_size_cells;
 
 	/* We are scanning "memory" nodes only */
 	if (type == NULL) {
@@ -912,11 +912,22 @@ int __init early_init_dt_scan_memory(unsigned long node, const char *uname,
 
 	pr_debug("memory scan node %s, reg size %d,\n", uname, l);
 
-	while ((endp - reg) >= (dt_root_addr_cells + dt_root_size_cells)) {
+
+	dt_mem_size_cells = dt_root_size_cells;
+	prop = of_get_flat_dt_prop(node, "#size-cells", NULL);
+	if (prop)
+		dt_mem_size_cells = be32_to_cpup(prop);
+
+	dt_mem_addr_cells = dt_root_addr_cells;
+	prop = of_get_flat_dt_prop(node, "#address-cells", NULL);
+	if (prop)
+		dt_mem_addr_cells = be32_to_cpup(prop);
+
+	while ((endp - reg) >= (dt_mem_addr_cells + dt_mem_size_cells)) {
 		u64 base, size;
 
-		base = dt_mem_next_cell(dt_root_addr_cells, &reg);
-		size = dt_mem_next_cell(dt_root_size_cells, &reg);
+		base = dt_mem_next_cell(dt_mem_addr_cells, &reg);
+		size = dt_mem_next_cell(dt_mem_size_cells, &reg);
 
 		if (size == 0)
 			continue;
-- 
1.9.1

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [next] | [standalone]

#1237369

On Thursday 01 October 2015 17:48:21 Vineet Gupta wrote:
> Currently memory node parsing uses root "#size-cells", "#address-cells"
> This doesn't work correctly when memory address/size is different or
> greater than root's.
> 
> e.g. ARC 32-bit systems implementing physical adressing extension and
> say 4GB of memory. All peripherals mappings stay within the 4GB (so root
> address/size cells remain 1 each), only the memory node address/size
> cells needs to specify greater than 32-bits as below
> 
>         memory {
>                 device_type = "memory";
>                 reg = <0x0 0x80000000 0x1 0x00000000>;  /* 4 GB */
>                 #address-cells = <2>;
>                 #size-cells = <2>;
>         };
> 
> This patch lets me boot a ARC system with PAE40 + 4GB of memory specified
> as above and fails to boot otherwise as memory parsing doesn't populate
> right base, size.
> 

This looks wrong: the #address-cells property in a device node is used
to parse the reg property of its child nodes, not the node itself.

The only way to list memory like this is to put #size-cells=<2>
into the root node. All lower bus nodes can then use the
normal #address-cells/#size-cells again and use a ranges property
to convert the register ranges so you don't need to update all
nodes.

	Arnd
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [next] | [standalone]

#1238766 — Re: [PATCH v2] of/fdt: Allow memory node and root to have different size/address cells

On Thursday 01 October 2015 05:52 PM, Arnd Bergmann wrote:
> On Thursday 01 October 2015 17:48:21 Vineet Gupta wrote:
>> Currently memory node parsing uses root "#size-cells", "#address-cells"
>> This doesn't work correctly when memory address/size is different or
>> greater than root's.
>>
>> e.g. ARC 32-bit systems implementing physical adressing extension and
>> say 4GB of memory. All peripherals mappings stay within the 4GB (so root
>> address/size cells remain 1 each), only the memory node address/size
>> cells needs to specify greater than 32-bits as below
>>
>>         memory {
>>                 device_type = "memory";
>>                 reg = <0x0 0x80000000 0x1 0x00000000>;  /* 4 GB */
>>                 #address-cells = <2>;
>>                 #size-cells = <2>;
>>         };
>>
>> This patch lets me boot a ARC system with PAE40 + 4GB of memory specified
>> as above and fails to boot otherwise as memory parsing doesn't populate
>> right base, size.
>>
> This looks wrong: the #address-cells property in a device node is used
> to parse the reg property of its child nodes, not the node itself.
>
> The only way to list memory like this is to put #size-cells=<2>
> into the root node. All lower bus nodes can then use the
> normal #address-cells/#size-cells again and use a ranges property
> to convert the register ranges so you don't need to update all
> nodes.

Thx for taking a look Arnd. I changed the DT per your suggestion and that works
equally well.

Thx,
-Vineet

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web