Re: [PATCH net-next] ptp: clockmatrix: use rsmu driver to access i2c/spi bus

From: Jakub Kicinski <kuba@kernel.org>
To: Min Li <min.li.xe@renesas.com>
Cc: "richardcochran@gmail.com" <richardcochran@gmail.com>,
	"netdev@vger.kernel.org" <netdev@vger.kernel.org>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"lee.jones@linaro.org" <lee.jones@linaro.org>
Subject: Re: [PATCH net-next] ptp: clockmatrix: use rsmu driver to access i2c/spi bus
Date: Thu, 23 Sep 2021 13:04:58 -0700	[thread overview]
Message-ID: <20210923130458.23cd40f2@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com> (raw)
In-Reply-To: <TYCPR01MB66086FB0DF1CBCC00F716A7ABAA39@TYCPR01MB6608.jpnprd01.prod.outlook.com>

On Thu, 23 Sep 2021 19:49:56 +0000 Min Li wrote:
> > > I did build it through 32 bit arm and didn't get the problem.
> > >
> > > make ARCH=arm CROSS_COMPILE=arm-linux-gnueabi-  
> > 
> > We're testing x86, maybe arm32 can handle 64bit divisions natively?
> > 
> > ERROR: modpost: "__divdi3" [drivers/ptp/ptp_clockmatrix.ko] undefined!
> > ERROR: modpost: "__udivdi3" [drivers/ptp/ptp_clockmatrix.ko] undefined!  
> 
> Hi Jakub
> 
> I tried "make ARCH=i386" but it also passed on my machine. Can you tell me how to
> reproduce this? Thanks

Hm, are you sure the config does not have CONFIG_64BIT=y ? Getting this
right can be tricky, here is the script which patchwork runs, FWIW:

https://github.com/kuba-moo/nipa/blob/master/tests/patch/build_32bit/build_32bit.sh

There is a small chance it's a glitch in the test system, but seems
unlikely, this looks like a 64b divide:

+static u32 idtcm_get_dco_delay(struct idtcm_channel *channel)
...
+	u64 m;
+	u16 n;
...
+	fodFreq = m / n;
...
+		return 18 * (u64)NSEC_PER_SEC / fodFreq;