Skip to contentSkip to conversation

fdt: A couple of no-map fixes

PATCH Applied Quentin Perret 3 participants 6 messages Jan 15, 2021 – Jan 16, 2021 Device tree

lore
Share & tools
lore
#0 lore raw

cover fdt: A couple of no-map fixes

In the context of a KVM/arm64 thread [1], it became apparent that the difference in behaviour of EFI and DT with respect to no-map regions was something worth fixing.

Rob pointed out in that thread that a couple of patches in this area had been sent but still needed to be integrated together, which is basically what this series attempts to do.

- Patch 01 is a generic fix that should make sense on its own;

- Patch 02 allows the kernel to have a saner behaviour with a broken DT. This one had a small rebase conflict (caused by patch 01). I tested it in qemu and everything is looking good, the error message shows up in dmesg when it should.

Thanks,
Quentin

[1] https://lore.kernel.org/kvmarm/[email protected]/

KarimAllah Ahmed (1):
fdt: Properly handle "no-map" field in the memory region

Nicolas Boichat (1):
of/fdt: Make sure no-map does not remove already reserved regions

 drivers/of/fdt.c | 12 ++++++++++--
 1 file changed, 10 insertions(+), 2 deletions(-)
Signature · 1 line
2.30.0.284.gd98b1dd5eaa7-goog
#1 lore raw

PATCH 1/2 fdt: Properly handle "no-map" field in the memory region

From: KarimAllah Ahmed <[email protected]>

Mark the memory region with NOMAP flag instead of completely removing it from the memory blocks. That makes the FDT handling consistent with the EFI memory map handling.

---
 drivers/of/fdt.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/of/fdt.c +1 -1
index feb0f2d67fc5..427b534d60d2 100644
@@ -1147,7 +1147,7 @@ int __init __weak early_init_dt_reserve_memory_arch(phys_addr_t base,
 					phys_addr_t size, bool nomap)
 {
 	if (nomap)
-		return memblock_remove(base, size);
+		return memblock_mark_nomap(base, size);
 	return memblock_reserve(base, size);
 }
 
Signature · 1 line
2.30.0.284.gd98b1dd5eaa7-goog
maintainer Applied
#2 lore raw

Re: [PATCH 1/2] fdt: Properly handle "no-map" field in the memory region

Quoted 12 lines · Quentin Perret

From: KarimAllah Ahmed <[email protected]>

Mark the memory region with NOMAP flag instead of completely removing it from the memory blocks. That makes the FDT handling consistent with the EFI memory map handling.

Quoted 4 lines
---
 drivers/of/fdt.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Applied, thanks!

#3 lore raw

PATCH 2/2 of/fdt: Make sure no-map does not remove already reserved regions

In reply to Quentin Perret

From: Nicolas Boichat <[email protected]>

If the device tree is incorrectly configured, and attempts to define a "no-map" reserved memory that overlaps with the kernel data/code, the kernel would crash quickly after boot, with no obvious clue about the nature of the issue.

For example, this would happen if we have the kernel mapped at these addresses (from /proc/iomem):
40000000-41ffffff : System RAM
40080000-40dfffff : Kernel code
40e00000-411fffff : reserved
41200000-413e0fff : Kernel data

And we declare a no-map shared-dma-pool region at a fixed address within that range:

mem_reserved: mem_region {
	compatible = "shared-dma-pool";
	reg = <0 0x40000000 0 0x01A00000>;
	no-map;
};

To fix this, when removing memory regions at early boot (which is what "no-map" regions do), we need to make sure that the memory is not already reserved. If we do, __reserved_mem_reserve_reg will throw an error:

[    0.000000] OF: fdt: Reserved memory: failed to reserve memory
   for node 'mem_region': base 0x0000000040000000, size 26 MiB

and the code that will try to use the region should also fail, later on.

We do not do anything for non-"no-map" regions, as memblock explicitly allows reserved regions to overlap, and the commit that this fixes removed the check for that precise reason.

[ qperret: fixed conflicts caused by the usage of memblock_mark_nomap ]

Fixes 094cb98179f19b7 ("of/fdt: memblock_reserve /memreserve/ regions in the case of partial overlap")
---
 drivers/of/fdt.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)
drivers/of/fdt.c +9 -1
index 427b534d60d2..dcc1dd96911a 100644
@@ -1146,8 +1146,16 @@ int __init __weak early_init_dt_mark_hotplug_memory_arch(u64 base, u64 size)
 int __init __weak early_init_dt_reserve_memory_arch(phys_addr_t base,
 					phys_addr_t size, bool nomap)
 {
-	if (nomap)
+	if (nomap) {
+		/*
+		 * If the memory is already reserved (by another region), we
+		 * should not allow it to be marked nomap.
+		 */
+		if (memblock_is_region_reserved(base, size))
+			return -EBUSY;
+
 		return memblock_mark_nomap(base, size);
+	}
 	return memblock_reserve(base, size);
 }
 
Signature · 1 line
2.30.0.284.gd98b1dd5eaa7-goog
maintainer Applied
#4 lore raw

Re: [PATCH 2/2] of/fdt: Make sure no-map does not remove already reserved regions

Quoted 37 lines · Quentin Perret

From: Nicolas Boichat <[email protected]>

If the device tree is incorrectly configured, and attempts to define a "no-map" reserved memory that overlaps with the kernel data/code, the kernel would crash quickly after boot, with no obvious clue about the nature of the issue.

For example, this would happen if we have the kernel mapped at these addresses (from /proc/iomem):
40000000-41ffffff : System RAM
40080000-40dfffff : Kernel code
40e00000-411fffff : reserved
41200000-413e0fff : Kernel data

And we declare a no-map shared-dma-pool region at a fixed address within that range:

mem_reserved: mem_region {
	compatible = "shared-dma-pool";
	reg = <0 0x40000000 0 0x01A00000>;
	no-map;
};

To fix this, when removing memory regions at early boot (which is what "no-map" regions do), we need to make sure that the memory is not already reserved. If we do, __reserved_mem_reserve_reg will throw an error:

[    0.000000] OF: fdt: Reserved memory: failed to reserve memory
   for node 'mem_region': base 0x0000000040000000, size 26 MiB

and the code that will try to use the region should also fail, later on.

We do not do anything for non-"no-map" regions, as memblock explicitly allows reserved regions to overlap, and the commit that this fixes removed the check for that precise reason.

[ qperret: fixed conflicts caused by the usage of memblock_mark_nomap ]

Quoted 8 lines
Fixes 094cb98179f19b7 ("of/fdt: memblock_reserve /memreserve/ regions in the case of partial overlap")
---
 drivers/of/fdt.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

Applied, thanks!

#5 lore raw

Re: [PATCH 2/2] of/fdt: Make sure no-map does not remove already reserved regions

Quoted 39 lines · Rob Herring
Quoted 37 lines · Quentin Perret

From: Nicolas Boichat <[email protected]>

If the device tree is incorrectly configured, and attempts to define a "no-map" reserved memory that overlaps with the kernel data/code, the kernel would crash quickly after boot, with no obvious clue about the nature of the issue.

For example, this would happen if we have the kernel mapped at these addresses (from /proc/iomem):
40000000-41ffffff : System RAM
40080000-40dfffff : Kernel code
40e00000-411fffff : reserved
41200000-413e0fff : Kernel data

And we declare a no-map shared-dma-pool region at a fixed address within that range:

mem_reserved: mem_region {
      compatible = "shared-dma-pool";
      reg = <0 0x40000000 0 0x01A00000>;
      no-map;
};

To fix this, when removing memory regions at early boot (which is what "no-map" regions do), we need to make sure that the memory is not already reserved. If we do, __reserved_mem_reserve_reg will throw an error:

[    0.000000] OF: fdt: Reserved memory: failed to reserve memory
   for node 'mem_region': base 0x0000000040000000, size 26 MiB

and the code that will try to use the region should also fail, later on.

We do not do anything for non-"no-map" regions, as memblock explicitly allows reserved regions to overlap, and the commit that this fixes removed the check for that precise reason.

[ qperret: fixed conflicts caused by the usage of memblock_mark_nomap ]

Quoted 8 lines
Quoted 8 lines
Fixes 094cb98179f19b7 ("of/fdt: memblock_reserve /memreserve/ regions in the case of partial overlap")
---
 drivers/of/fdt.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

Thanks for picking up this old patch Quentin ,-)

Quoted 1 line

Applied, thanks!