netdev - Re: [PATCH net-next v3 01/12] net: wwan: t7xx: Add control DMA interface

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, 12 Jan 2022 21:24:59 +0200 (EET)
From:   Ilpo Järvinen <ilpo.jarvinen@...ux.intel.com>
To:     "Martinez, Ricardo" <ricardo.martinez@...ux.intel.com>
cc:     Andy Shevchenko <andriy.shevchenko@...ux.intel.com>,
        Netdev <netdev@...r.kernel.org>, linux-wireless@...r.kernel.org,
        kuba@...nel.org, davem@...emloft.net, johannes@...solutions.net,
        ryazanov.s.a@...il.com, loic.poulain@...aro.org,
        m.chetan.kumar@...el.com, chandrashekar.devegowda@...el.com,
        linuxwwan@...el.com, chiranjeevi.rapolu@...ux.intel.com,
        haijun.liu@...iatek.com, amir.hanania@...el.com,
        dinesh.sharma@...el.com, eliot.lee@...el.com,
        mika.westerberg@...ux.intel.com, moises.veleta@...el.com,
        pierre-louis.bossart@...el.com, muralidharan.sethuraman@...el.com,
        Soumya.Prakash.Mishra@...el.com, sreehari.kancharla@...el.com
Subject: Re: [PATCH net-next v3 01/12] net: wwan: t7xx: Add control DMA
 interface

On Wed, 12 Jan 2022, Martinez, Ricardo wrote:

> 
> On 1/12/2022 10:16 AM, Ilpo Järvinen wrote:
> > On Wed, 12 Jan 2022, Andy Shevchenko wrote:
> > 
> > > On Wed, Jan 12, 2022 at 04:24:52PM +0200, Ilpo Järvinen wrote:
> > > > On Wed, 12 Jan 2022, Andy Shevchenko wrote:
> > > > > On Tue, Jan 11, 2022 at 08:55:58PM -0800, Martinez, Ricardo wrote:
> > > > > > On 12/16/2021 3:08 AM, Ilpo Järvinen wrote:
> > > > > > > On Mon, 6 Dec 2021, Ricardo Martinez wrote:
> > > > > > > > +	if (req->entry.next == &ring->gpd_ring)
> > > > > > > > +		return list_first_entry(&ring->gpd_ring, struct
> > > > > > > > cldma_request, entry);
> > > > > > > > +
> > > > > > > > +	return list_next_entry(req, entry);
> > > > > ...
> > > > > 
> > > > > > > > +	if (req->entry.prev == &ring->gpd_ring)
> > > > > > > > +		return list_last_entry(&ring->gpd_ring, struct
> > > > > > > > cldma_request, entry);
> > > > > > > > +
> > > > > > > > +	return list_prev_entry(req, entry);
> > > > > ...
> > > > > 
> > > > > > > Wouldn't these two seems generic enough to warrant adding
> > > > > > > something like
> > > > > > > list_next/prev_entry_circular(...) to list.h?
> > > > > > Agree, in the upcoming version I'm planning to include something
> > > > > > like this
> > > > > > to list.h as suggested:
> > > > > I think you mean for next and prev, i.o.w. two helpers, correct?
> > > > > 
> > > > > > #define list_next_entry_circular(pos, ptr, member) \
> > One thing I missed earlier, the sigrature should instead of ptr have head:
> > #define list_next_entry_circular(pos, head, member)
> > 
> > > > > >      ((pos)->member.next == (ptr) ? \
> > > > > I believe this is list_entry_is_head().
> > > > It takes .next so it's not the same as list_entry_is_head() and
> > > > list_entry_is_last() doesn't exist.
> > > But we have list_last_entry(). So, what about
> > > 
> > > list_last_entry() == pos ? first : next;
> > > 
> > > and counterpart
> > > 
> > > list_first_entry() == pos ? last : prev;
> > > 
> > > ?
> > Yes, although now that I think it more, using them implies the head
> > element has to be always accessed. It might be marginally cache friendlier
> > to use list_entry_is_head you originally suggested but get the next entry
> > first:
> > ({
> > 	typeof(pos) next__ = list_next_entry(pos, member); \
> > 	!list_entry_is_head(next__, head, member) ? next__ :
> > list_next_entry(next__, member);
> > })
> > (This was written directly to email, entirely untested).
> > 
> > Here, the head element would only get accessed when we really need to walk
> > through it.
> 
> I'm not sure if list_next_entry() will work for the last element, what about
> using list_is_last()?

Why wouldn't it? E.g., list_for_each_entry() does it for the last entry 
before terminating the for loop.

-- 
 i.

> This way we avoid accessing head if not needed and does not to use
> 'container_of()' on (pos)->member.next.
> 
>     (list_is_last(&(pos)->member, head) ? \
>     list_first_entry(head, typeof(*(pos)), member) : \
>     list_next_entry(pos, member))
> 
> (untested)
> 
> > > > > >      list_first_entry(ptr, typeof(*(pos)), member) : \
> > > > > >      list_next_entry(pos, member))
>