[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20200618185507.p5tmurllwygmfwrq@ltop.local>
Date: Thu, 18 Jun 2020 20:55:07 +0200
From: Luc Van Oostenryck <luc.vanoostenryck@...il.com>
To: Linus Torvalds <torvalds@...ux-foundation.org>
Cc: kernel test robot <lkp@...el.com>, kbuild-all@...ts.01.org,
LKML <linux-kernel@...r.kernel.org>,
Miguel Ojeda <miguel.ojeda.sandonis@...il.com>
Subject: Re: [PATCH] sparse: use identifiers to define address spaces
On Thu, Jun 18, 2020 at 09:59:05AM -0700, Linus Torvalds wrote:
>
> The new warnings don't seem to be due to the kernel test robot having
> an old version of sparse, but just because the error strings changed,
> and presumably the kernel test robot has some "ignore old sparse
> warnings" logic.
>
> So the warnings all look new, even if they aren't.
What is strange is that only some of these warnings are shown as
new, but not all. For example it shows
>> drivers/spi/spi-fsl-spi.c:93:44: sparse: got void [noderef] __iomem *reg_base
as new, but not the following one:
drivers/spi/spi-fsl-spi.c:294:44: sparse: got void [noderef] __iomem *reg_base
while both have their string changed (__iomem).
Just to be sure, I checked this file and (of course) all these
warnings are old ones, caused by a missing __iomem in the assignments:
struct fsl_spi_reg *reg_base = mspi->reg_base;
Why some are shown as new and some as old is a mystery to me.
-- Luc
Powered by blists - more mailing lists