[PATCH] agp: check if ndevs is zero before enabling node

[PATCH] agp: check if ndevs is zero before enabling node

[Yiyuan GUO]

In function agp_3_5_enable from drivers/char/agp/isoch.c, the
variable ndevs may remain zero if all AGP devices have type of
"Bridge" or "Unclassified device". Passing ndevs==0 to function
 agp_3_5_isochronous_node_enable or agp_3_5_nonisochronous_node_enable
will lead to divide by zero problems.

Signed-off-by: Yiyuan GUO <yguoaz@cse.ust.hk>
---
 drivers/char/agp/isoch.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/drivers/char/agp/isoch.c b/drivers/char/agp/isoch.c
index 7ecf20a6d..4c21bb057 100644
--- a/drivers/char/agp/isoch.c
+++ b/drivers/char/agp/isoch.c
@@ -433,6 +433,11 @@ int agp_3_5_enable(struct agp_bridge_data *bridge)
 		}
 	}		
 
+	if (!ndevs) {
+		ret = -ENODEV;
+		goto free_and_exit;
+	}
+
 	/*
 	 * Call functions to divide target resources amongst the AGP 3.0
 	 * masters.  This process is dramatically different depending on
-- 
2.25.1

Re: [PATCH] agp: check if ndevs is zero before enabling node

[Greg KH]

On Thu, May 13, 2021 at 09:53:32PM +0800, Yiyuan GUO wrote:
> In function agp_3_5_enable from drivers/char/agp/isoch.c, the
> variable ndevs may remain zero if all AGP devices have type of
> "Bridge" or "Unclassified device".

Can that actually happen with real hardware?  Given the age of AGP code,
odds are if this could have happened, it would have by now :)

thanks,

greg k-h

Re: Re: [PATCH] agp: check if ndevs is zero before enabling node

[Yiyuan GUO]

From: Yiyuan GUO <yguoaz@gmail.com>

I guess it may only happen in theory, but maybe there is no harm in 
adding a simple check?

Thanks.