linux-kernel - Re: [PATCH v7 0/2] mm/vmalloc: lock contention optimization under multi-threading

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

Date: Wed, 6 Mar 2024 17:18:04 +0800
From: "Huang, Rulin" <rulin.huang@...el.com>
To: urezki@...il.com, bhe@...hat.com
Cc: akpm@...ux-foundation.org, colin.king@...el.com, hch@...radead.org,
 linux-kernel@...r.kernel.org, linux-mm@...ck.org, lstoakes@...il.com,
 tianyou.li@...el.com, tim.c.chen@...el.com, wangyang.guo@...el.com,
 zhiguo.zhou@...el.com, rulin.huang@...el.com
Subject: Re: [PATCH v7 0/2] mm/vmalloc: lock contention optimization under
 multi-threading

Hello, are there any issues with this patch that need to be modified? If
there is any, we will modify it as soon as possible, thank you.

On 2024/3/1 23:54, rulinhuang wrote:
> Hi,
> 
> This version has the rearrangement of macros from the previous one.
> 
> We are not sure whether we have completely moved these macros and 
> their corresponding helper to the correct position. Could you please 
> help to check whether they are correct?
> 
> ~
> 
> 1. Motivation
> 
> When allocating a new memory area where the mapping address range is 
> known, it is observed that the vmap_node->busy.lock is acquired twice 
> but one of the acquisitions is actually unnecessary.
> 
> 2. Design
> 
> Among the two acquisitions, the first one occurs in the 
> alloc_vmap_area() function when inserting the vm area into the vm 
> mapping red-black tree, and the second one occurs in the 
> setup_vmalloc_vm() function when updating the properties of the vm, 
> such as flags and address, etc.
> 
> Combine these two operations together in alloc_vmap_area(), which 
> improves scalability when the vmap_node->busy.lock is contended.
> By doing so, the need to acquire the lock twice can also be eliminated 
> to once.
> 
> 3. Test results
> 
> With the above change, tested on intel sapphire rapids
> platform(224 vcpu), a 4% performance improvement is gained on 
> stress-ng/pthread(https://github.com/ColinIanKing/stress-ng),
> which is the stress test of thread creations.
> 
> rulinhuang
> 
> [v1] https://lore.kernel.org/all/20240207033059.1565623-1-rulin.huang@intel.com/
> [v2] https://lore.kernel.org/all/20240220090521.3316345-1-rulin.huang@intel.com/
> [v3] https://lore.kernel.org/all/20240221032905.11392-1-rulin.huang@intel.com/
> [v4] https://lore.kernel.org/all/20240222120536.216166-1-rulin.huang@intel.com/
> [v5] https://lore.kernel.org/all/20240223130318.112198-2-rulin.huang@intel.com/
> [v6] https://lore.kernel.org/lkml/aa8f0413-d055-4b49-bcd3-401e93e01c6d@intel.com/
> 
> 
> rulinhuang (2):
>   mm/vmalloc: Moved macros with no functional change happened
>   mm/vmalloc: Eliminated the lock contention from twice to once
> 
>  mm/vmalloc.c | 314 +++++++++++++++++++++++++--------------------------
>  1 file changed, 155 insertions(+), 159 deletions(-)
> 
> 
> base-commit: 10c2cf5fe97647d68ee89b1f921e982e71519f20