[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-Id: <20230622183613.58762-1-andriy.shevchenko@linux.intel.com>
Date: Thu, 22 Jun 2023 21:36:10 +0300
From: Andy Shevchenko <andriy.shevchenko@...ux.intel.com>
To: Mark Brown <broonie@...nel.org>,
Andy Shevchenko <andriy.shevchenko@...ux.intel.com>,
linux-kernel@...r.kernel.org
Cc: Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
"Rafael J. Wysocki" <rafael@...nel.org>,
Linus Walleij <linus.walleij@...aro.org>
Subject: [PATCH v1 0/3] regmap: Drop never (properly) worked 64-bit support
regmap API internally operates on unsigned int values for the register
offsets and data. The commit back in 2015 that introduces 64-bit
excerpts in the code made a false impression that it works. Not really.
Consider two things:
1/ register offset
2/ data
For the first one is very rarely we need (except probably an MMIO case)
it. Even though, it won't work due to 32-bit limitations of the base offset.
Considering, let's say, 4 bytes stride the current implementation may
cover 36-bit of address space _only_. And 37-bit for the 8 bytes stride.
For the second one it's obviously that we want _all_ bits to be covered
in the data (otherwise what's the point?) and unsigned int gives us
only 32-bits.
With all this, revert all 64-bit excerpts from regmap API to avoid
false impressions and new code that never works.
Note, there are no users with such sizes in the kernel.
Andy Shevchenko (3):
regmap: Revert "add 64-bit mode support" and Co.
regmap: cache: Revert "Add 64-bit mode support"
regmap: mmio: Remove unused 64-bit support code
drivers/base/regmap/regcache.c | 15 ----
drivers/base/regmap/regmap-mmio.c | 24 ------
drivers/base/regmap/regmap.c | 122 ------------------------------
3 files changed, 161 deletions(-)
--
2.40.0.1.gaa8946217a0b
Powered by blists - more mailing lists