Message-ID: <20230706170213.2ea63606@thinkpad-T15>
Date:   Thu, 6 Jul 2023 17:02:13 +0200
From:   Gerald Schaefer <gerald.schaefer@...ux.ibm.com>
To:     Hugh Dickins <hughd@...gle.com>
Cc:     Jason Gunthorpe <jgg@...pe.ca>,
        Andrew Morton <akpm@...ux-foundation.org>,
        Vasily Gorbik <gor@...ux.ibm.com>,
        Mike Kravetz <mike.kravetz@...cle.com>,
        Mike Rapoport <rppt@...nel.org>,
        "Kirill A. Shutemov" <kirill.shutemov@...ux.intel.com>,
        Matthew Wilcox <willy@...radead.org>,
        David Hildenbrand <david@...hat.com>,
        Suren Baghdasaryan <surenb@...gle.com>,
        Qi Zheng <zhengqi.arch@...edance.com>,
        Yang Shi <shy828301@...il.com>,
        Mel Gorman <mgorman@...hsingularity.net>,
        Peter Xu <peterx@...hat.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Will Deacon <will@...nel.org>, Yu Zhao <yuzhao@...gle.com>,
        Alistair Popple <apopple@...dia.com>,
        Ralph Campbell <rcampbell@...dia.com>,
        Ira Weiny <ira.weiny@...el.com>,
        Steven Price <steven.price@....com>,
        SeongJae Park <sj@...nel.org>,
        Lorenzo Stoakes <lstoakes@...il.com>,
        Huang Ying <ying.huang@...el.com>,
        Naoya Horiguchi <naoya.horiguchi@....com>,
        Christophe Leroy <christophe.leroy@...roup.eu>,
        Zack Rusin <zackr@...are.com>,
        Axel Rasmussen <axelrasmussen@...gle.com>,
        Anshuman Khandual <anshuman.khandual@....com>,
        Pasha Tatashin <pasha.tatashin@...een.com>,
        Miaohe Lin <linmiaohe@...wei.com>,
        Minchan Kim <minchan@...nel.org>,
        Christoph Hellwig <hch@...radead.org>,
        Song Liu <song@...nel.org>,
        Thomas Hellstrom <thomas.hellstrom@...ux.intel.com>,
        Russell King <linux@...linux.org.uk>,
        "David S. Miller" <davem@...emloft.net>,
        Michael Ellerman <mpe@...erman.id.au>,
        "Aneesh Kumar K.V" <aneesh.kumar@...ux.ibm.com>,
        Heiko Carstens <hca@...ux.ibm.com>,
        Christian Borntraeger <borntraeger@...ux.ibm.com>,
        Claudio Imbrenda <imbrenda@...ux.ibm.com>,
        Alexander Gordeev <agordeev@...ux.ibm.com>,
        Jann Horn <jannh@...gle.com>,
        Vishal Moola <vishal.moola@...il.com>,
        Vlastimil Babka <vbabka@...e.cz>,
        linux-arm-kernel@...ts.infradead.org, sparclinux@...r.kernel.org,
        linuxppc-dev@...ts.ozlabs.org, linux-s390@...r.kernel.org,
        linux-kernel@...r.kernel.org, linux-mm@...ck.org
Subject: Re: [PATCH v2 07/12] s390: add pte_free_defer() for pgtables
 sharing page

On Wed, 5 Jul 2023 18:20:21 -0700 (PDT)
Hugh Dickins <hughd@...gle.com> wrote:

> On Wed, 5 Jul 2023, Gerald Schaefer wrote:
> > On Tue, 4 Jul 2023 10:03:57 -0700 (PDT)
> > Hugh Dickins <hughd@...gle.com> wrote:  
> > > On Tue, 4 Jul 2023, Gerald Schaefer wrote:  
> > > > On Sat, 1 Jul 2023 21:32:38 -0700 (PDT)
> > > > Hugh Dickins <hughd@...gle.com> wrote:    
> > > > > On Thu, 29 Jun 2023, Hugh Dickins wrote:    
> > > ...  
> > > > > --- a/arch/s390/mm/pgalloc.c
> > > > > +++ b/arch/s390/mm/pgalloc.c
> > > > > @@ -229,6 +229,15 @@ void page_table_free_pgste(struct page *page)
> > > > >   * logic described above. Both AA bits are set to 1 to denote a 4KB-pgtable
> > > > >   * while the PP bits are never used, nor such a page is added to or removed
> > > > >   * from mm_context_t::pgtable_list.
> > > > > + *
> > > > > + * pte_free_defer() overrides those rules: it takes the page off pgtable_list,
> > > > > + * and prevents both 2K fragments from being reused. pte_free_defer() has to
> > > > > + * guarantee that its pgtable cannot be reused before the RCU grace period
> > > > > + * has elapsed (which page_table_free_rcu() does not actually guarantee).    
> > > > 
> > > > Hmm, I think page_table_free_rcu() has to guarantee the same, i.e. not
> > > > allow reuse before grace period elapsed. And I hope that it does so, by
> > > > setting the PP bits, which would be noticed in page_table_alloc(), in
> > > > case the page would be seen there.
> > > > 
> > > > Unlike pte_free_defer(), page_table_free_rcu() would add pages back to the
> > > > end of the list, and so they could be seen in page_table_alloc(), but they
> > > > should not be reused before grace period elapsed and __tlb_remove_table()
> > > > cleared the PP bits, as far as I understand.
> > > > 
> > > > So what exactly do you mean with "which page_table_free_rcu() does not actually
> > > > guarantee"?    
> > > 
> > > I'll answer without locating and re-reading what Jason explained earlier,
> > > perhaps in a separate thread, about pseudo-RCU-ness in tlb_remove_table():
> > > he may have explained it better.  And without working out again all the
> > > MMU_GATHER #defines, and which of them do and do not apply to s390 here.
> > > 
> > > The detail that sticks in my mind is the fallback in tlb_remove_table()  
> > 
> > Ah ok, I was aware of that "semi-RCU" fallback logic in tlb_remove_table(),
> > but that is rather a generic issue, and not s390-specific.  
> 
> Yes.
> 
> > I thought you
> > meant some s390-oddity here, of which we have a lot, unfortunately...
> > Of course, we call tlb_remove_table() from our page_table_free_rcu(), so
> > I guess you could say that page_table_free_rcu() cannot guarantee what
> > tlb_remove_table() cannot guarantee.
> > 
> > Maybe change to "which page_table_free_rcu() does not actually guarantee,
> > by calling tlb_remove_table()", to make it clear that this is not a problem
> > of page_table_free_rcu() itself.  
> 
> Okay - I'll rephrase slightly to avoid being sued by s390's lawyers :-)
> 
> >   
> > > in mm/mmu_gather.c: if its __get_free_page(GFP_NOWAIT) fails, it cannot
> > > batch the tables for freeing by RCU, and resorts instead to an immediate 
> > > TLB flush (I think: that again involves chasing definitions) followed by
> > > tlb_remove_table_sync_one() - which just delivers an interrupt to each CPU,
> > > and is commented: 
> > > /*
> > >  * This isn't an RCU grace period and hence the page-tables cannot be
> > >  * assumed to be actually RCU-freed.
> > >  *
> > >  * It is however sufficient for software page-table walkers that rely on
> > >  * IRQ disabling.
> > >  */
> > > 
> > > Whether that's good for your PP pages or not, I've given no thought:
> > > I've just taken it on trust that what s390 has working today is good.  
> > 
> > Yes, we should be fine with that, current code can be trusted :-)  
> 
> Glad to hear it :-)  Yes, I think it's not actually relying on the "rcu"
> implied by the function name.

Ah ok, now I get it. Never noticed that naming it "free_rcu" could be
misleading. It is only ever called from pte_free_tlb(), so always in that
"semi-RCU" context. If you just look at the name, you could expect this
to always free pagetables by RCU, which would be exactly what you need
for pte_free_defer(), and which of course cannot be guaranteed by our
page_table_free_rcu().

IOW, exactly what your comment says, and now I think it is actually fine
as it is :-)

I guess I am a bit lamebrained this week, due to early shift and not
enough sleep...

> 
> >   
> > > 
> > > If that __get_free_page(GFP_NOWAIT) fallback instead used call_rcu(),
> > > then I would not have written "(which page_table_free_rcu() does not
> > > actually guarantee)".  But it cannot use call_rcu() because it does
> > > not have an rcu_head to work with - it's in some generic code, and
> > > there is no MMU_GATHER_CAN_USE_PAGE_RCU_HEAD for architectures to set.
> > > 
> > > And Jason would have much preferred us to address the issue from that
> > > angle; but not only would doing so destroy my sanity, I'd also destroy
> > > 20 architectures TLB-flushing, unbuilt and untested, in the attempt.  
> > 
> > Oh yes, if your changes would have allowed to get rid of that "semi RCU"
> > logic, that would really be a major boost in popularity, I guess. But
> > it probably is as it is, because it is not so easily fixed...  
> 
> I'm hoping that this series might help stir someone else to get into that.
> 
> >   
> > > 
> > > ...  
> > > > > @@ -325,10 +346,17 @@ void page_table_free(struct mm_struct *mm, unsigned long *table)
> > > > >  		 */
> > > > >  		mask = atomic_xor_bits(&page->_refcount, 0x11U << (bit + 24));
> > > > >  		mask >>= 24;
> > > > > -		if (mask & 0x03U)
> > > > > +		if ((mask & 0x03U) && !PageActive(page)) {
> > > > > +			/*
> > > > > +			 * Other half is allocated, and neither half has had
> > > > > +			 * its free deferred: add page to head of list, to make
> > > > > +			 * this freed half available for immediate reuse.
> > > > > +			 */
> > > > >  			list_add(&page->lru, &mm->context.pgtable_list);
> > > > > -		else
> > > > > -			list_del(&page->lru);
> > > > > +		} else {
> > > > > +			/* If page is on list, now remove it. */
> > > > > +			list_del_init(&page->lru);
> > > > > +		}    
> > > > 
> > > > Ok, we might end up with some unnecessary list_del_init() here, e.g. if
> > > > other half is still allocated, when called from pte_free_defer() on a
> > > > fully allocated page, which was not on the list (and with PageActive, and
> > > > (mask & 0x03U) true).
> > > > Not sure if adding an additional mask check to the else path would be
> > > > needed, but it seems that list_del_init() should also be able to handle
> > > > this.    
> > > 
> > > list_del_init() is very cheap in the unnecessary case: the cachelines
> > > required are already there.  You don't want a flag to say whether to
> > > call it or not, it is already the efficient approach.  
> > 
> > Yes, I also see no functional issue here. Just thought that the extra
> > write could be avoided, e.g. by checking for list_empty() or mask first.
> > But I guess that is simply the benefit of list_del_init(), that you
> > don't have to check, at least if it is guaranteed that rcu_head is
> > never in use here.
> > 
> > Then maybe adjust the comment, because now it makes you wonder, when
> > you read (and understand) the code, you see that this list_del_init()
> > might also be called for pages not on the list.  
> 
> Sorry, I don't understand what clarification you're asking for there.
> I thought
> 			/* If page is on list, now remove it. */
> 			list_del_init(&page->lru);
> was good enough comment.
> 
> (I certainly don't want to enumerate the cases when it is or is not
> already on the list there, that would be misery; but I don't think
> that's the adjustment you were asking for either.)

I was mislead by the comment saying "If page is on the list", in an
else path where we also end up for pages not on the list any more.
I guess I would have added something like "it is also ok to do
list_del_init() here for pages not on the list". But thinking again,
that would probably just be a reminder of how list_del_init() works, 
which should be obvious anyway, at least for people with enough sleep.

> 
> >   
> > > 
> > > (But you were right not to use it in your pt_frag_refcount version,
> > > because there we were still trying to do the call_rcu() per fragment
> > > rather than per page, so page->lru could have been on the RCU queue.)  
> > 
> > That is actually the one thing I still try to figure out, by drawing
> > pictures, i.e. if we really really never end up here on list_del_init(),
> > while using rcu_head, e.g. by racing PageActive.  
> 
> There is no race with PageActive being seen when the table page is
> finally to be freed (by RCU or not).  But there is definitely a harmless
> race with pte_free_defer()er of other half setting PageActive an instant
> after page_table_free() checked PageActive here.  So maybe this
> page_table_free() does a list_add(), which the racer then list_del_init()s
> when it gets the mm->context.lock; or maybe they both list_del_init().

Agree.

Since none of my remarks on the comments seem valid or strictly necessary
any more, and I also could not find functional issues, I think you can add
this patch as new version for 07/12. And I can now give you this:

Reviewed-by: Gerald Schaefer <gerald.schaefer@...ux.ibm.com>

Powered by blists - more mailing lists