2023-03-29 16:54:39

by Denis Plotnikov

Subject: [PATCH] net: netxen: report error on version offset reading

A static analyzer complains for non-checking the function returning value.
Although, the code looks like not expecting any problems with version
reading on netxen_p3_has_mn call, it seems the error still may happen.
So, at least, add error reporting to ease problems investigation.

Signed-off-by: Denis Plotnikov <[email protected]>
---
drivers/net/ethernet/qlogic/netxen/netxen_nic_init.c | 9 +++++++--
1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/qlogic/netxen/netxen_nic_init.c b/drivers/net/ethernet/qlogic/netxen/netxen_nic_init.c
index 35ec9aab3dc7b..92962dbb73ad0 100644
--- a/drivers/net/ethernet/qlogic/netxen/netxen_nic_init.c
+++ b/drivers/net/ethernet/qlogic/netxen/netxen_nic_init.c
@@ -1192,8 +1192,13 @@ netxen_p3_has_mn(struct netxen_adapter *adapter)
if (NX_IS_REVISION_P2(adapter->ahw.revision_id))
return 1;

- netxen_rom_fast_read(adapter,
- NX_FW_VERSION_OFFSET, (int *)&flashed_ver);
+ if (netxen_rom_fast_read(adapter,
+ NX_FW_VERSION_OFFSET, (int *)&flashed_ver)) {
+ printk(KERN_ERR "%s: ERROR on flashed version reading",
+ netxen_nic_driver_name);
+ return 0;
+ }
+
flashed_ver = NETXEN_DECODE_VERSION(flashed_ver);

if (flashed_ver >= NETXEN_VERSION_CODE(4, 0, 220)) {
--
2.25.1

2023-03-29 19:04:04

by Leon Romanovsky

Subject: Re: [PATCH] net: netxen: report error on version offset reading

On Wed, Mar 29, 2023 at 07:26:29PM +0300, Denis Plotnikov wrote:
> A static analyzer complains for non-checking the function returning value.
> Although, the code looks like not expecting any problems with version
> reading on netxen_p3_has_mn call, it seems the error still may happen.
> So, at least, add error reporting to ease problems investigation.
>
> Signed-off-by: Denis Plotnikov <[email protected]>
> ---
> drivers/net/ethernet/qlogic/netxen/netxen_nic_init.c | 9 +++++++--
> 1 file changed, 7 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/net/ethernet/qlogic/netxen/netxen_nic_init.c b/drivers/net/ethernet/qlogic/netxen/netxen_nic_init.c
> index 35ec9aab3dc7b..92962dbb73ad0 100644
> --- a/drivers/net/ethernet/qlogic/netxen/netxen_nic_init.c
> +++ b/drivers/net/ethernet/qlogic/netxen/netxen_nic_init.c
> @@ -1192,8 +1192,13 @@ netxen_p3_has_mn(struct netxen_adapter *adapter)
> if (NX_IS_REVISION_P2(adapter->ahw.revision_id))
> return 1;
>
> - netxen_rom_fast_read(adapter,
> - NX_FW_VERSION_OFFSET, (int *)&flashed_ver);
> + if (netxen_rom_fast_read(adapter,
> + NX_FW_VERSION_OFFSET, (int *)&flashed_ver)) {

1. Mo callers of netxen_rom_fast_read() print debug messages, so this
shouldn't too.
2. netxen_p3_has_mn() can't fail and by returning 0, you will cause to
unpredictable behaviour in netxen_validate_firmware().

Thanks

> + printk(KERN_ERR "%s: ERROR on flashed version reading",
> + netxen_nic_driver_name);
> + return 0;
> + }
> +
> flashed_ver = NETXEN_DECODE_VERSION(flashed_ver);
>
> if (flashed_ver >= NETXEN_VERSION_CODE(4, 0, 220)) {
> --
> 2.25.1
>

2023-03-30 06:03:50

by Denis Plotnikov

Subject: Re: [PATCH] net: netxen: report error on version offset reading

On 29.03.2023 21:59, Leon Romanovsky wrote:
> On Wed, Mar 29, 2023 at 07:26:29PM +0300, Denis Plotnikov wrote:
>> A static analyzer complains for non-checking the function returning value.
>> Although, the code looks like not expecting any problems with version
>> reading on netxen_p3_has_mn call, it seems the error still may happen.
>> So, at least, add error reporting to ease problems investigation.
>>
>> Signed-off-by: Denis Plotnikov <[email protected]>
>> ---
>> drivers/net/ethernet/qlogic/netxen/netxen_nic_init.c | 9 +++++++--
>> 1 file changed, 7 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/net/ethernet/qlogic/netxen/netxen_nic_init.c b/drivers/net/ethernet/qlogic/netxen/netxen_nic_init.c
>> index 35ec9aab3dc7b..92962dbb73ad0 100644
>> --- a/drivers/net/ethernet/qlogic/netxen/netxen_nic_init.c
>> +++ b/drivers/net/ethernet/qlogic/netxen/netxen_nic_init.c
>> @@ -1192,8 +1192,13 @@ netxen_p3_has_mn(struct netxen_adapter *adapter)
>> if (NX_IS_REVISION_P2(adapter->ahw.revision_id))
>> return 1;
>>
>> - netxen_rom_fast_read(adapter,
>> - NX_FW_VERSION_OFFSET, (int *)&flashed_ver);
>> + if (netxen_rom_fast_read(adapter,
>> + NX_FW_VERSION_OFFSET, (int *)&flashed_ver)) {
> 1. Mo callers of netxen_rom_fast_read() print debug messages, so this
> shouldn't too.
> 2. netxen_p3_has_mn() can't fail and by returning 0, you will cause to
> unpredictable behaviour in netxen_validate_firmware().
>
> Thanks

Well, ok. Then patch isn't needed.

Thanks for reviewing!


Denis

>
>> + printk(KERN_ERR "%s: ERROR on flashed version reading",
>> + netxen_nic_driver_name);
>> + return 0;
>> + }
>> +
>> flashed_ver = NETXEN_DECODE_VERSION(flashed_ver);
>>
>> if (flashed_ver >= NETXEN_VERSION_CODE(4, 0, 220)) {
>> --
>> 2.25.1
>>