linux-kernel - RE: [PATCH] scsi: ufs: core: increase the NOP

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]

Message-ID: <351601db67ba$b67a54d0$236efe70$@samsung.com>
Date: Thu, 16 Jan 2025 11:02:35 +0900
From: DooHyun Hwang(황두현/Device S/W Solution Lab.(MX)/삼성전자)
	<dh0421.hwang@...sung.com>
To: "'Bart Van Assche'" <bvanassche@....org>, <linux-scsi@...r.kernel.org>,
	<linux-kernel@...r.kernel.org>, <alim.akhtar@...sung.com>,
	<avri.altman@....com>, <James.Bottomley@...senPartnership.com>,
	<martin.petersen@...cle.com>, <peter.wang@...iatek.com>,
	<manivannan.sadhasivam@...aro.org>, <quic_mnaresh@...cinc.com>
Cc: <grant.jung@...sung.com>, <jt77.jang@...sung.com>,
	<junwoo80.lee@...sung.com>, <jangsub.yi@...sung.com>,
	<sh043.lee@...sung.com>, <cw9316.lee@...sung.com>,
	<sh8267.baek@...sung.com>, <wkon.kim@...sung.com>
Subject: RE: [PATCH] scsi: ufs: core: increase the NOP_OUT command timeout

> On 1/14/25 6:23 PM, DooHyun Hwang wrote:
> > It is found that is UFS device may take longer than 500ms(50ms *
> > 10times) to respond to NOP_OUT command.
> >
> > The NOP_OUT command timeout was total 500ms that is from a timeout
> > value of 50ms(defined by NOP_OUT_TIMEOUT) with 10 retries(defined by
> > NOP_OUT_RETRIES)
> >
> > This change increase the NOP_OUT command timeout to total 1000ms by
> > changing timeout value to 100ms(NOP_OUT_TIMEOUT)
> >
> > Signed-off-by: DooHyun Hwang <dh0421.hwang@...sung.com>
> > ---
> >   drivers/ufs/core/ufshcd.c | 4 ++--
> >   1 file changed, 2 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
> > index cd404ade48dc..bf5c4620ef6b 100644
> > --- a/drivers/ufs/core/ufshcd.c
> > +++ b/drivers/ufs/core/ufshcd.c
> > @@ -57,8 +57,8 @@ enum {
> >   };
> >   /* NOP OUT retries waiting for NOP IN response */
> >   #define NOP_OUT_RETRIES    10
> > -/* Timeout after 50 msecs if NOP OUT hangs without response */
> > -#define NOP_OUT_TIMEOUT    50 /* msecs */
> > +/* Timeout after 100 msecs if NOP OUT hangs without response */
> > +#define NOP_OUT_TIMEOUT    100 /* msecs */
> >
> >   /* Query request retries */
> >   #define QUERY_REQ_RETRIES 3
> 
> The above change relies on all device management commands being issued
> with the same tag. If a single NOP OUT command may take longer than
> 500 ms, shouldn't NOP_OUT_TIMEOUT be increased to 1000 ms instead of
> 100 ms? The number of NOP OUT retries seems high to me and probably can be
> reduced?
> 
> Thanks,
> 
> Bart.
I want to keep sending NOP_OUT commands repeatedly to get a response
from the UFS device, as per the existing method. To accommodate this method,
I propose increasing the total timeout duration by increasing the single timeout
value(defined by NOP_OUT_TIMEOUT) from 50ms to 100ms rather than
increasing the timeout value(defined by NOP_OUT_TIMEOUT) from 50ms to 1000ms
or increasing the retry count value(defined by NOP_OUT_RETRIES).

This is time measurement log confirmed on a real device with NOP_OUT_TIMEOUT is 100ms

1. normal operation
[    2.010156] [6:  kworker/u18:0:   76] ufshcd-qcom 1d84000.ufshc: [TEST] ufshcd_verify_dev_init: takes 1271 us, retries = 1 * 100ms.

2. issued log : exceeds 500ms
[    2.524525] [6:  kworker/u17:2:  141] ufshcd-qcom 1d84000.ufshc: [TEST] ufshcd_verify_dev_init: takes 533000 us, retries = 6 * 100ms.

And a certain UFS vendor has confirmed that the response to NOP_OUT command
can be delayed by up to 540ms in certain circumstances on a specific model.

BR,
Thank you.
DooHyun Hwang.