[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20200330132028.GA22483@bombadil.infradead.org>
Date: Mon, 30 Mar 2020 06:20:28 -0700
From: Matthew Wilcox <willy@...radead.org>
To: Wei Yang <richard.weiyang@...il.com>
Cc: linux-fsdevel@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 2/9] XArray: simplify the calculation of shift
On Mon, Mar 30, 2020 at 12:36:36PM +0000, Wei Yang wrote:
> When head is NULL, shift is calculated from max. Currently we use a loop
> to detect how many XA_CHUNK_SHIFT is need to cover max.
>
> To achieve this, we can get number of bits max expands and round it up
> to XA_CHUNK_SHIFT.
>
> Signed-off-by: Wei Yang <richard.weiyang@...il.com>
> ---
> lib/xarray.c | 6 +-----
> 1 file changed, 1 insertion(+), 5 deletions(-)
>
> diff --git a/lib/xarray.c b/lib/xarray.c
> index 1d9fab7db8da..6454cf3f5b4c 100644
> --- a/lib/xarray.c
> +++ b/lib/xarray.c
> @@ -560,11 +560,7 @@ static int xas_expand(struct xa_state *xas, void *head)
> unsigned long max = xas_max(xas);
>
> if (!head) {
> - if (max == 0)
> - return 0;
> - while ((max >> shift) >= XA_CHUNK_SIZE)
> - shift += XA_CHUNK_SHIFT;
> - return shift + XA_CHUNK_SHIFT;
> + return roundup(fls_long(max), XA_CHUNK_SHIFT);
This doesn't give the same number. Did you test this?
Consider max = 64. The current code does:
shift = 0;
64 >> 0 >= 64 (true)
shift += 6;
64 >> 6 < 64
return 12
Your replacement does:
fls_long(64) = 6
roundup(6, 6) is 6.
Please be more careful.
Powered by blists - more mailing lists