RE: [EXTERNAL] Re: [PATCH v5] i2c: designware: Enable SAR in slave mode for Multiple SAR controllers
From: Umesh Kumar
Date: Tue Aug 18 2026 - 10:19:22 EST
Hi Andy ,
Apologies for the oversight — no functional changes were made since your Reviewed-by.
I inadvertently dropped the tag during commit amend. Resending as v6 with your Reviewed-by included.
Regards ,
Umesh Kumar
From: Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx>
Sent: Monday, August 17, 2026 10:04 PM
To: Umesh Kumar <ukumar@xxxxxxxxxxx>
Cc: Andi Shyti <andi.shyti@xxxxxxxxxx>; Mika Westerberg <mika.westerberg@xxxxxxxxxxxxxxx>; Bharat Bhushan <bbhushan2@xxxxxxxxxxx>; linux-i2c@xxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx
Subject: [EXTERNAL] Re: [PATCH v5] i2c: designware: Enable SAR in slave mode for Multiple SAR controllers
On Mon, Aug 17, 2026 at 09: 51: 41PM +0530, Umesh Kumar wrote: > DesignWare controllers synthesised with the Multiple SAR feature (e. g. > Marvell CN20K) gate the primary slave address behind IC_ENABLE. SAR_EN > (bit 19). Per the Synopsys
ZjQcmQRYFpfptBannerStart
Prioritize security for external emails:
Confirm sender and content safety before clicking links or opening attachments
https://us-phishalarm-ewt.proofpoint.com/EWT/v1/CRVmXkqW!tO3TvhD0whzUOE920RVUQ_XJPV3L26S-lEMnq92RJA9M7hx481kZx8jKjFzdQ9FYO_HrR3ywL34PqSFqWA$ ;
ZjQcmQRYFpfptBannerEnd
On Mon, Aug 17, 2026 at 09:51:41PM +0530, Umesh Kumar wrote:
> DesignWare controllers synthesised with the Multiple SAR feature (e.g.
> Marvell CN20K) gate the primary slave address behind IC_ENABLE.SAR_EN
> (bit 19). Per the Synopsys DW_apb_i2c DataBook v2.03a (Multiple SAR
> feature), when this bit is set the core acknowledges the address matching
> IC_SAR; otherwise the address is NACKed. The slave enable path writes only
> IC_ENABLE.ENABLE (bit 0), so on such controllers the target never ACKs
> its address and a remote master cannot reach it.
>
> Set SAR_EN together with ENABLE when entering slave mode. The bit is
> reserved (and ignored) on single-SAR controllers, so existing users are
> unaffected.
I gave my tag and it is absent here. What's going on?
Do you think the fix is not only that and requires a new round of review?
--
With Best Regards,
Andy Shevchenko