2019-06-17 13:05:53

by Arnd Bergmann

[permalink] [raw]
Subject: [PATCH] qed: Fix -Wmaybe-uninitialized false positive

A previous attempt to shut up the uninitialized variable use
warning was apparently insufficient. When CONFIG_PROFILE_ANNOTATED_BRANCHES
is set, gcc-8 still warns, because the unlikely() check in DP_NOTICE()
causes it to no longer track the state of all variables correctly:

drivers/net/ethernet/qlogic/qed/qed_dev.c: In function 'qed_llh_set_ppfid_affinity':
drivers/net/ethernet/qlogic/qed/qed_dev.c:798:47: error: 'abs_ppfid' may be used uninitialized in this function [-Werror=maybe-uninitialized]
addr = NIG_REG_PPF_TO_ENGINE_SEL + abs_ppfid * 0x4;
~~~~~~~~~~^~~~~

This is not a nice workaround, but always initializing the output from
qed_llh_abs_ppfid() at least shuts up the false positive reliably.

Fixes: 79284adeb99e ("qed: Add llh ppfid interface and 100g support for offload protocols")
Fixes: 8e2ea3ea9625 ("qed: Fix static checker warning")
Signed-off-by: Arnd Bergmann <[email protected]>
---
drivers/net/ethernet/qlogic/qed/qed_dev.c | 1 +
1 file changed, 1 insertion(+)

diff --git a/drivers/net/ethernet/qlogic/qed/qed_dev.c b/drivers/net/ethernet/qlogic/qed/qed_dev.c
index eec7cb65c7e6..a1ebc2b1ca0b 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_dev.c
+++ b/drivers/net/ethernet/qlogic/qed/qed_dev.c
@@ -652,6 +652,7 @@ static int qed_llh_abs_ppfid(struct qed_dev *cdev, u8 ppfid, u8 *p_abs_ppfid)
DP_NOTICE(cdev,
"ppfid %d is not valid, available indices are 0..%hhd\n",
ppfid, p_llh_info->num_ppfid - 1);
+ *p_abs_ppfid = 0;
return -EINVAL;
}

--
2.20.0


2019-06-18 09:29:20

by Michal Kalderon

[permalink] [raw]
Subject: RE: [PATCH] qed: Fix -Wmaybe-uninitialized false positive

> From: Arnd Bergmann <[email protected]>
> Sent: Monday, June 17, 2019 4:05 PM
>
> A previous attempt to shut up the uninitialized variable use warning was
> apparently insufficient. When CONFIG_PROFILE_ANNOTATED_BRANCHES is
> set, gcc-8 still warns, because the unlikely() check in DP_NOTICE() causes it to
> no longer track the state of all variables correctly:
>
> drivers/net/ethernet/qlogic/qed/qed_dev.c: In function
> 'qed_llh_set_ppfid_affinity':
> drivers/net/ethernet/qlogic/qed/qed_dev.c:798:47: error: 'abs_ppfid' may
> be used uninitialized in this function [-Werror=maybe-uninitialized]
> addr = NIG_REG_PPF_TO_ENGINE_SEL + abs_ppfid * 0x4;
> ~~~~~~~~~~^~~~~
>
> This is not a nice workaround, but always initializing the output from
> qed_llh_abs_ppfid() at least shuts up the false positive reliably.
>
> Fixes: 79284adeb99e ("qed: Add llh ppfid interface and 100g support for
> offload protocols")
> Fixes: 8e2ea3ea9625 ("qed: Fix static checker warning")
> Signed-off-by: Arnd Bergmann <[email protected]>
> ---
> drivers/net/ethernet/qlogic/qed/qed_dev.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/drivers/net/ethernet/qlogic/qed/qed_dev.c
> b/drivers/net/ethernet/qlogic/qed/qed_dev.c
> index eec7cb65c7e6..a1ebc2b1ca0b 100644
> --- a/drivers/net/ethernet/qlogic/qed/qed_dev.c
> +++ b/drivers/net/ethernet/qlogic/qed/qed_dev.c
> @@ -652,6 +652,7 @@ static int qed_llh_abs_ppfid(struct qed_dev *cdev, u8
> ppfid, u8 *p_abs_ppfid)
> DP_NOTICE(cdev,
> "ppfid %d is not valid, available indices are
> 0..%hhd\n",
> ppfid, p_llh_info->num_ppfid - 1);
> + *p_abs_ppfid = 0;
> return -EINVAL;
> }
>
> --
> 2.20.0

Thanks,?

Acked-by: Michal Kalderon?<[email protected]>


2019-06-18 17:45:19

by David Miller

[permalink] [raw]
Subject: Re: [PATCH] qed: Fix -Wmaybe-uninitialized false positive

From: Arnd Bergmann <[email protected]>
Date: Mon, 17 Jun 2019 15:04:49 +0200

> A previous attempt to shut up the uninitialized variable use
> warning was apparently insufficient. When CONFIG_PROFILE_ANNOTATED_BRANCHES
> is set, gcc-8 still warns, because the unlikely() check in DP_NOTICE()
> causes it to no longer track the state of all variables correctly:
>
> drivers/net/ethernet/qlogic/qed/qed_dev.c: In function 'qed_llh_set_ppfid_affinity':
> drivers/net/ethernet/qlogic/qed/qed_dev.c:798:47: error: 'abs_ppfid' may be used uninitialized in this function [-Werror=maybe-uninitialized]
> addr = NIG_REG_PPF_TO_ENGINE_SEL + abs_ppfid * 0x4;
> ~~~~~~~~~~^~~~~
>
> This is not a nice workaround, but always initializing the output from
> qed_llh_abs_ppfid() at least shuts up the false positive reliably.
>
> Fixes: 79284adeb99e ("qed: Add llh ppfid interface and 100g support for offload protocols")
> Fixes: 8e2ea3ea9625 ("qed: Fix static checker warning")
> Signed-off-by: Arnd Bergmann <[email protected]>

Applied to net-next.