linux-kernel - Re: [net PATCH v2] octeontx2-af: Unlock contexts in the queue context cache in case of fault detection

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Thu, 23 Feb 2023 14:16:37 +0100
From:   Simon Horman <simon.horman@...igine.com>
To:     Sai Krishna <saikrishnag@...vell.com>
Cc:     davem@...emloft.net, edumazet@...gle.com, kuba@...nel.org,
        pabeni@...hat.com, netdev@...r.kernel.org,
        linux-kernel@...r.kernel.org, sgoutham@...vell.com,
        sumang@...vell.com
Subject: Re: [net PATCH v2] octeontx2-af: Unlock contexts in the queue
 context cache in case of fault detection

On Thu, Feb 23, 2023 at 04:31:25PM +0530, Sai Krishna wrote:
> From: Suman Ghosh <sumang@...vell.com>
> 
> NDC caches contexts of frequently used queue's (Rx and Tx queues)
> contexts. Due to a HW errata when NDC detects fault/poision while
> accessing contexts it could go into an illegal state where a cache
> line could get locked forever. To makesure all cache lines in NDC
> are available for optimum performance upon fault/lockerror/posion
> errors scan through all cache lines in NDC and clear the lock bit.
> 
> Fixes: 4a3581cd5995 ("octeontx2-af: NPA AQ instruction enqueue support")
> Signed-off-by: Suman Ghosh <sumang@...vell.com>
> Signed-off-by: Sunil Kovvuri Goutham <sgoutham@...vell.com>
> Signed-off-by: Sai Krishna <saikrishnag@...vell.com>

...

> diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu.h b/drivers/net/ethernet/marvell/octeontx2/af/rvu.h
> index 389663a13d1d..6508f25b2b37 100644
> --- a/drivers/net/ethernet/marvell/octeontx2/af/rvu.h
> +++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu.h
> @@ -884,6 +884,12 @@ int rvu_cpt_lf_teardown(struct rvu *rvu, u16 pcifunc, int blkaddr, int lf,
>  int rvu_cpt_ctx_flush(struct rvu *rvu, u16 pcifunc);
>  int rvu_cpt_init(struct rvu *rvu);
>  
> +/* NDC APIs */
> +#define NDC_MAX_BANK(rvu, blk_addr) (rvu_read64(rvu, \
> +					blk_addr, NDC_AF_CONST) & 0xFF)
> +#define NDC_MAX_LINE_PER_BANK(rvu, blk_addr) ((rvu_read64(rvu, \
> +					blk_addr, NDC_AF_CONST) & 0xFFFF0000) >> 16)

Perhaps not appropriate to include as part of a fix,
as NDC_MAX_BANK is being moved from elsewhere,
but I wonder if this might be more cleanly implemented
using FIELD_GET().

...

> diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_reg.h b/drivers/net/ethernet/marvell/octeontx2/af/rvu_reg.h
> index 1729b22580ce..bc6ca5ccc1ff 100644
> --- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_reg.h
> +++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_reg.h
> @@ -694,6 +694,7 @@
>  #define NDC_AF_INTR_ENA_W1S		(0x00068)
>  #define NDC_AF_INTR_ENA_W1C		(0x00070)
>  #define NDC_AF_ACTIVE_PC		(0x00078)
> +#define NDC_AF_CAMS_RD_INTERVAL		(0x00080)
>  #define NDC_AF_BP_TEST_ENABLE		(0x001F8)
>  #define NDC_AF_BP_TEST(a)		(0x00200 | (a) << 3)
>  #define NDC_AF_BLK_RST			(0x002F0)
> @@ -709,6 +710,8 @@
>  		(0x00F00 | (a) << 5 | (b) << 4)
>  #define NDC_AF_BANKX_HIT_PC(a)		(0x01000 | (a) << 3)
>  #define NDC_AF_BANKX_MISS_PC(a)		(0x01100 | (a) << 3)
> +#define NDC_AF_BANKX_LINEX_METADATA(a, b) \
> +		(0x10000 | (a) << 3 | (b) << 3)

It looks a little odd that both a and b are shifted by 3 bits.
If it's intended then perhaps it would be clearer to write this as:

#define NDC_AF_BANKX_LINEX_METADATA(a, b) \
		(0x10000 | ((a) | (b)) << 3)

>  
>  /* LBK */
>  #define LBK_CONST			(0x10ull)
> -- 
> 2.25.1
>