The current check of nvec < minvec for nvec returned from
platform_irq_count() will not detect a negative error code in nvec.
This is because minvec is unsigned, and, as such, nvec is promoted to
unsigned in that check, which will make it a huge number (if it contained
-EPROBE_DEFER).
In practice, an error should not occur in nvec for the only in-tree
user, but add a check anyway.
Fixes: e15f2fa959f2 ("driver core: platform: Add
devm_platform_get_irqs_affinity()")
Reported-by: Dan Carpenter <dan.carpen...@oracle.com>
Signed-off-by: John Garry <john.ga...@huawei.com>
---
I hope that this can go through either irqchip or driver/core trees, thanks!
diff --git a/drivers/base/platform.c b/drivers/base/platform.c
index 95fd1549f87d..8456d8384ac8 100644
--- a/drivers/base/platform.c
+++ b/drivers/base/platform.c
@@ -366,6 +366,8 @@ int devm_platform_get_irqs_affinity(struct platform_device
*dev,
return -ERANGE;
nvec = platform_irq_count(dev);
+ if (nvec < 0)
+ return nvec;
if (nvec < minvec)
return -ENOSPC;
--
2.26.2
