linux-kernel - [PATCH 00/13] lib/generic-radix-tree: genradix bug fix and optimisations.

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-ID: <21289d79b0474811b21ed8478c465159@AcuMS.aculab.com>
Date:   Tue, 25 Aug 2020 14:52:34 +0000
From:   David Laight <David.Laight@...LAB.COM>
To:     "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "'linux-sctp@...r.kernel.org'" <linux-sctp@...r.kernel.org>,
        Eric Biggers <ebiggers@...nel.org>,
        'Marcelo Ricardo Leitner' <marcelo.leitner@...il.com>,
        'Catalin Marinas' <catalin.marinas@....com>,
        "'kent.overstreet@...il.com'" <kent.overstreet@...il.com>,
        Andrew Morton <akpm@...ux-foundation.org>,
        "'Neil Horman'" <nhorman@...driver.com>
Subject: [PATCH 00/13] lib/generic-radix-tree: genradix bug fix and
 optimisations.

The genradix code is used by SCTP for accessing per-stream data.
This means there are quite a lot of lookups but the code wasn't
really optimised at all.

Patch 1 fixes a rather nasty bug that could cause all sorts of oddities
including having a loop in a tree.
Fortunately the only 2 users of the genradix code in the kernel source
tree and neither can possibly hit the bug.

Patches 2 through 11 all reduce codepath. Especially for the common
case (for both users) where all the data fits in one page.

Patch 12 inlines the common lookup function for 1 page trees.
This also remove the '% constant' calcuation from the lookups.
However it will increase code size (but not codepath) because
of the number of callers.

Patch 13 removes part of the optimisation for 1 page tress
under the assumption that the lookup will have been inlined.
The code still works even if inlining didn't happen.

David Laight (13):
    01/13: Fix potentially corrupt tree
    02/13: Optimise out ilog2(variable).
    03/13: Always use low 8 bits of 'root' for depth.
    04/13: Optimise __genradix_ptr()
    05/13: Optimise __genradix_ptr_alloc()
    06/13: Rename gfp_mask to gfp to shorten lines.
    07/13: Optimise __genradix_iter_peek()
    08/13: Save number of bits to shift instead of tree level.
    09/13: Check sizeof(_type) when defining a tree.
    10/13: Simplify offset calculation:
    11/13: Pass the root pointer to __genradix_ptr.
    12/13: Inline genradix_ptr() for simple trees.
    13/13: Simplify __genradix_ptr()

 include/linux/generic-radix-tree.h |  54 ++++---
 lib/generic-radix-tree.c           | 224 ++++++++++++++++-------------
 2 files changed, 162 insertions(+), 116 deletions(-)

-- 
2.25.1

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)