The struct resource end field is inclusive not exclusive, this is, the
size is (end - start) +1.
Update the definitions and use the generic resource_size() function.
Fixes cocci check:
drivers/media/pci/mgb4/mgb4_regs.c:13:22-25: WARNING: Suspicious code. resource_size is maybe missing with res
Reviewed-by: Martin Tůma <[email protected]>
Signed-off-by: Ricardo Ribalda <[email protected]>
---
drivers/media/pci/mgb4/mgb4_core.c | 4 ++--
drivers/media/pci/mgb4/mgb4_regs.c | 2 +-
2 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/drivers/media/pci/mgb4/mgb4_core.c b/drivers/media/pci/mgb4/mgb4_core.c
index 9bcf10a77fd3..60498a5abebf 100644
--- a/drivers/media/pci/mgb4/mgb4_core.c
+++ b/drivers/media/pci/mgb4/mgb4_core.c
@@ -493,13 +493,13 @@ static int mgb4_probe(struct pci_dev *pdev, const struct pci_device_id *id)
struct mgb4_dev *mgbdev;
struct resource video = {
.start = 0x0,
- .end = 0x100,
+ .end = 0xff,
.flags = IORESOURCE_MEM,
.name = "mgb4-video",
};
struct resource cmt = {
.start = 0x1000,
- .end = 0x1800,
+ .end = 0x17ff,
.flags = IORESOURCE_MEM,
.name = "mgb4-cmt",
};
diff --git a/drivers/media/pci/mgb4/mgb4_regs.c b/drivers/media/pci/mgb4/mgb4_regs.c
index 53d4e4503a74..31befd722d72 100644
--- a/drivers/media/pci/mgb4/mgb4_regs.c
+++ b/drivers/media/pci/mgb4/mgb4_regs.c
@@ -10,7 +10,7 @@
int mgb4_regs_map(struct resource *res, struct mgb4_regs *regs)
{
regs->mapbase = res->start;
- regs->mapsize = res->end - res->start;
+ regs->mapsize = resource_size(res);
if (!request_mem_region(regs->mapbase, regs->mapsize, res->name))
return -EINVAL;
--
2.44.0.769.g3c40516874-goog
On 19/04/2024 10:47, Ricardo Ribalda wrote:
> The struct resource end field is inclusive not exclusive, this is, the
> size is (end - start) +1.
", this is," doesn't parse on my end
"i.e" => that is, would be more appropriate I think.
"The struct resource end field is inclusive not exclusive of the size"
which I still think is a confusing statement.
Perhaps something much easier to understand is called for
"The struct resource end field signifies the end address not the
relative offset from the start field i.e size == (end - start) + 1.
Amend the .end field to specify the end address not the relative size
from the offset as is currently given."
Other than that, I think its reasonable to assume the mapping != 0 -
0x100 inclusive.
Please consider updating your commit log and if you do add my
Reviewed-by: Bryan O'Donoghue <[email protected]>
---
bod