Re: [PATCH stable v5.15.y 1/1] crypto: add __init/__exit annotations to init/exit funcs

[Greg KH]

On Wed, Feb 15, 2023 at 11:10:14PM +0000, Saeed Mirzamohammadi wrote:
> Hi,
> 
> > On Feb 14, 2023, at 10:55 PM, Greg KH <gregkh@xxxxxxxxxxxxxxxxxxx> wrote:
> > 
> > On Tue, Feb 14, 2023 at 11:53:00AM -0800, Saeed Mirzamohammadi wrote:
> >> From: Xiu Jianfeng <xiujianfeng@xxxxxxxxxx>
> >> 
> >> Add missing __init/__exit annotations to init/exit funcs.
> >> 
> >> Signed-off-by: Xiu Jianfeng <xiujianfeng@xxxxxxxxxx>
> >> Signed-off-by: Herbert Xu <herbert@xxxxxxxxxxxxxxxxxxx>
> >> (cherry picked from commit 33837be33367172d66d1f2bd6964cc41448e6e7c)
> >> Cc: stable@xxxxxxxxxxxxxxx # 5.15+
> >> Signed-off-by: Saeed Mirzamohammadi <saeed.mirzamohammadi@xxxxxxxxxx>
> >> ---
> >> crypto/async_tx/raid6test.c | 4 ++--
> >> crypto/curve25519-generic.c | 4 ++--
> >> crypto/dh.c                 | 4 ++--
> >> crypto/ecdh.c               | 4 ++--
> >> crypto/ecdsa.c              | 4 ++--
> >> crypto/rsa.c                | 4 ++--
> >> crypto/sm2.c                | 4 ++--
> >> 7 files changed, 14 insertions(+), 14 deletions(-)
> > 
> > What bug/problem does this resolve?  Why is this needed in stable
> > kernels?
> 
> I don’t have any specific bug to discuss for this fix. It would help freeing up some memory after initialization for these crypto modules since the init function is only called once but not addressing any major issue. Feel free to disregard if this is trivial for stable.

Exactly how much memory is actually freed here and why is that required
for a stable kernel?

Please read the kernel documentation for what is considered a valid
stable kernel change:
    https://www.kernel.org/doc/html/latest/process/stable-kernel-rules.html

thanks,

greg k-h