Thread (3 messages) 3 messages, 2 authors, 2021-05-28

Re: [PATCH] crypto: arm64/gcm - remove Wunused-const-variable ghash_cpu_feature

From: Austin Kim <hidden>
Date: 2021-05-28 13:51:45
Also in: linux-arm-kernel, linux-crypto, lkml

2021년 5월 28일 (금) 오전 2:46, Eric Biggers [off-list ref]님이 작성:
On Thu, May 27, 2021 at 07:28:09AM +0100, Austin Kim wrote:
quoted
The variable with MODULE_DEVICE_TABLE() is registered as platform_driver.
What does this mean?  There is no platform_driver involved here at all.
What I would like to say is "in many cases, MODULE_DEVICE_TABLE is
registered as struct of platform_driver"
ex) [drivers/char/hw_random/meson-rng.c]
MODULE_DEVICE_TABLE(of, meson_rng_of_match);

static struct platform_driver meson_rng_driver = {
.probe = meson_rng_probe,
.driver = {
.name = "meson-rng",
.of_match_table = meson_rng_of_match,
},
};
quoted
But ghash_cpu_feature is not used, so remove ghash_cpu_feature.
It is used when the file is built as a module.
Thanks for the information. But it looks like this file is compiled
as built-in with 'CONFIG_CRYPTO_GHASH_ARM64_CE=y' by default.
quoted
diff --git a/arch/arm64/crypto/ghash-ce-glue.c b/arch/arm64/crypto/ghash-ce-glue.c
index 720cd3a58da3..c3f27d0d5329 100644
[...]
Probably adding __maybe_unused to ghash_cpu_feature[] is the right thing to do.
That's what module_cpu_feature_match() does.

(Note that module_cpu_feature_match() can't be used here, as it seems the intent
is for this module to be autoloaded when PMULL is detected, but still be
loadable without it.  So, that's apparently the reason for using
MODULE_DEVICE_TABLE() directly.)
Thanks for the valuable idea and feedback.
- Eric
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help