| Message ID | 20190722153209.73107-1-roger.pau@citrix.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | x86/p2m: fix non-translated handling of iommu mappings | expand |
> -----Original Message----- > From: Roger Pau Monne <roger.pau@citrix.com> > Sent: 22 July 2019 16:32 > To: xen-devel@lists.xenproject.org > Cc: Roger Pau Monne <roger.pau@citrix.com>; George Dunlap <George.Dunlap@citrix.com>; Jan Beulich > <jbeulich@suse.com>; Andrew Cooper <Andrew.Cooper3@citrix.com>; Wei Liu <wl@xen.org>; Paul Durrant > <Paul.Durrant@citrix.com> > Subject: [PATCH] x86/p2m: fix non-translated handling of iommu mappings > > The current usage of need_iommu_pt_sync in p2m for non-translated > guests is wrong because it doesn't correctly handle a relaxed PV > hardware domain, that has need_sync set to false, but still need > entries to be added from calls to {set/clear}_identity_p2m_entry. > > Adjust the code in guest_physmap_add_page to also check whether the > domain has an iommu instead of whether it needs syncing or not in > order to take a reference to a page to be mapped. > > Signed-off-by: Roger Pau Monné <roger.pau@citrix.com> > --- > Cc: George Dunlap <george.dunlap@eu.citrix.com> > Cc: Jan Beulich <jbeulich@suse.com> > Cc: Andrew Cooper <andrew.cooper3@citrix.com> > Cc: Wei Liu <wl@xen.org> > Cc: Paul Durrant <Paul.Durrant@citrix.com> Reviewed-by: Paul Durrant <paul.durrant@citrix.com> > --- > xen/arch/x86/mm/p2m.c | 6 +++--- > 1 file changed, 3 insertions(+), 3 deletions(-) > > diff --git a/xen/arch/x86/mm/p2m.c b/xen/arch/x86/mm/p2m.c > index fef97c82f6..88a2430c8c 100644 > --- a/xen/arch/x86/mm/p2m.c > +++ b/xen/arch/x86/mm/p2m.c > @@ -836,7 +836,7 @@ guest_physmap_add_page(struct domain *d, gfn_t gfn, mfn_t mfn, > */ > for ( i = 0; i < (1UL << page_order); ++i, ++page ) > { > - if ( !need_iommu_pt_sync(d) ) > + if ( !has_iommu_pt(d) ) > /* nothing */; > else if ( get_page_and_type(page, d, PGT_writable_page) ) > put_page_and_type(page); > @@ -1341,7 +1341,7 @@ int set_identity_p2m_entry(struct domain *d, unsigned long gfn_l, > > if ( !paging_mode_translate(p2m->domain) ) > { > - if ( !need_iommu_pt_sync(d) ) > + if ( !has_iommu_pt(d) ) > return 0; > return iommu_legacy_map(d, _dfn(gfn_l), _mfn(gfn_l), PAGE_ORDER_4K, > IOMMUF_readable | IOMMUF_writable); > @@ -1432,7 +1432,7 @@ int clear_identity_p2m_entry(struct domain *d, unsigned long gfn_l) > > if ( !paging_mode_translate(d) ) > { > - if ( !need_iommu_pt_sync(d) ) > + if ( !has_iommu_pt(d) ) > return 0; > return iommu_legacy_unmap(d, _dfn(gfn_l), PAGE_ORDER_4K); > } > -- > 2.20.1 (Apple Git-117)
On 22.07.2019 17:32, Roger Pau Monne wrote: > The current usage of need_iommu_pt_sync in p2m for non-translated > guests is wrong because it doesn't correctly handle a relaxed PV > hardware domain, that has need_sync set to false, but still need > entries to be added from calls to {set/clear}_identity_p2m_entry. > > Adjust the code in guest_physmap_add_page to also check whether the > domain has an iommu instead of whether it needs syncing or not in > order to take a reference to a page to be mapped. Why this seemingly unrelated change? I ask because ... > --- a/xen/arch/x86/mm/p2m.c > +++ b/xen/arch/x86/mm/p2m.c > @@ -836,7 +836,7 @@ guest_physmap_add_page(struct domain *d, gfn_t gfn, mfn_t mfn, > */ > for ( i = 0; i < (1UL << page_order); ++i, ++page ) > { > - if ( !need_iommu_pt_sync(d) ) > + if ( !has_iommu_pt(d) ) > /* nothing */; > else if ( get_page_and_type(page, d, PGT_writable_page) ) > put_page_and_type(page); ... this parallels the code in iommu_hwdom_init(), which similarly uses need_iommu_pt_sync() (and during the prior discussion you did agree with Paul that it shouldn't be changed). IOW the patch for now can have my R-b only with this change and the respective part of the description dropped. Jan
On Tue, Jul 23, 2019 at 10:32:41AM +0000, Jan Beulich wrote: > On 22.07.2019 17:32, Roger Pau Monne wrote: > > The current usage of need_iommu_pt_sync in p2m for non-translated > > guests is wrong because it doesn't correctly handle a relaxed PV > > hardware domain, that has need_sync set to false, but still need > > entries to be added from calls to {set/clear}_identity_p2m_entry. > > > > Adjust the code in guest_physmap_add_page to also check whether the > > domain has an iommu instead of whether it needs syncing or not in > > order to take a reference to a page to be mapped. > > Why this seemingly unrelated change? I ask because ... > > > --- a/xen/arch/x86/mm/p2m.c > > +++ b/xen/arch/x86/mm/p2m.c > > @@ -836,7 +836,7 @@ guest_physmap_add_page(struct domain *d, gfn_t gfn, mfn_t mfn, > > */ > > for ( i = 0; i < (1UL << page_order); ++i, ++page ) > > { > > - if ( !need_iommu_pt_sync(d) ) > > + if ( !has_iommu_pt(d) ) > > /* nothing */; > > else if ( get_page_and_type(page, d, PGT_writable_page) ) > > put_page_and_type(page); > > ... this parallels the code in iommu_hwdom_init(), which similarly > uses need_iommu_pt_sync() (and during the prior discussion you did > agree with Paul that it shouldn't be changed). IOW the patch for > now can have my R-b only with this change and the respective part > of the description dropped. OK, this is again not needed for a relaxed PV hardware domain because all RAM is already mapped on the iommu page tables. Let me update the commit and resend. Thanks, Roger.
On Mon, Jul 22, 2019 at 8:33 AM Roger Pau Monne <roger.pau@citrix.com> wrote: > > The current usage of need_iommu_pt_sync in p2m for non-translated > guests is wrong because it doesn't correctly handle a relaxed PV > hardware domain, that has need_sync set to false, but still need > entries to be added from calls to {set/clear}_identity_p2m_entry. > > Adjust the code in guest_physmap_add_page to also check whether the > domain has an iommu instead of whether it needs syncing or not in > order to take a reference to a page to be mapped. > > Signed-off-by: Roger Pau Monné <roger.pau@citrix.com> > --- > Cc: George Dunlap <george.dunlap@eu.citrix.com> > Cc: Jan Beulich <jbeulich@suse.com> > Cc: Andrew Cooper <andrew.cooper3@citrix.com> > Cc: Wei Liu <wl@xen.org> > Cc: Paul Durrant <Paul.Durrant@citrix.com> Tested-by: Roman Shaposhnik <roman@zededa.com> > --- > xen/arch/x86/mm/p2m.c | 6 +++--- > 1 file changed, 3 insertions(+), 3 deletions(-) > > diff --git a/xen/arch/x86/mm/p2m.c b/xen/arch/x86/mm/p2m.c > index fef97c82f6..88a2430c8c 100644 > --- a/xen/arch/x86/mm/p2m.c > +++ b/xen/arch/x86/mm/p2m.c > @@ -836,7 +836,7 @@ guest_physmap_add_page(struct domain *d, gfn_t gfn, mfn_t mfn, > */ > for ( i = 0; i < (1UL << page_order); ++i, ++page ) > { > - if ( !need_iommu_pt_sync(d) ) > + if ( !has_iommu_pt(d) ) > /* nothing */; > else if ( get_page_and_type(page, d, PGT_writable_page) ) > put_page_and_type(page); > @@ -1341,7 +1341,7 @@ int set_identity_p2m_entry(struct domain *d, unsigned long gfn_l, > > if ( !paging_mode_translate(p2m->domain) ) > { > - if ( !need_iommu_pt_sync(d) ) > + if ( !has_iommu_pt(d) ) > return 0; > return iommu_legacy_map(d, _dfn(gfn_l), _mfn(gfn_l), PAGE_ORDER_4K, > IOMMUF_readable | IOMMUF_writable); > @@ -1432,7 +1432,7 @@ int clear_identity_p2m_entry(struct domain *d, unsigned long gfn_l) > > if ( !paging_mode_translate(d) ) > { > - if ( !need_iommu_pt_sync(d) ) > + if ( !has_iommu_pt(d) ) > return 0; > return iommu_legacy_unmap(d, _dfn(gfn_l), PAGE_ORDER_4K); > } > -- > 2.20.1 (Apple Git-117) > > > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.xenproject.org > https://lists.xenproject.org/mailman/listinfo/xen-devel
On Mon, Aug 19, 2019 at 06:57:06PM -0700, Roman Shaposhnik wrote: > On Mon, Jul 22, 2019 at 8:33 AM Roger Pau Monne <roger.pau@citrix.com> wrote: > > > > The current usage of need_iommu_pt_sync in p2m for non-translated > > guests is wrong because it doesn't correctly handle a relaxed PV > > hardware domain, that has need_sync set to false, but still need > > entries to be added from calls to {set/clear}_identity_p2m_entry. > > > > Adjust the code in guest_physmap_add_page to also check whether the > > domain has an iommu instead of whether it needs syncing or not in > > order to take a reference to a page to be mapped. > > > > Signed-off-by: Roger Pau Monné <roger.pau@citrix.com> > > --- > > Cc: George Dunlap <george.dunlap@eu.citrix.com> > > Cc: Jan Beulich <jbeulich@suse.com> > > Cc: Andrew Cooper <andrew.cooper3@citrix.com> > > Cc: Wei Liu <wl@xen.org> > > Cc: Paul Durrant <Paul.Durrant@citrix.com> > > Tested-by: Roman Shaposhnik <roman@zededa.com> There's a v2 of the patch, but I guess your TB also applies to it. Thanks, Roger.
diff --git a/xen/arch/x86/mm/p2m.c b/xen/arch/x86/mm/p2m.c index fef97c82f6..88a2430c8c 100644 --- a/xen/arch/x86/mm/p2m.c +++ b/xen/arch/x86/mm/p2m.c @@ -836,7 +836,7 @@ guest_physmap_add_page(struct domain *d, gfn_t gfn, mfn_t mfn, */ for ( i = 0; i < (1UL << page_order); ++i, ++page ) { - if ( !need_iommu_pt_sync(d) ) + if ( !has_iommu_pt(d) ) /* nothing */; else if ( get_page_and_type(page, d, PGT_writable_page) ) put_page_and_type(page); @@ -1341,7 +1341,7 @@ int set_identity_p2m_entry(struct domain *d, unsigned long gfn_l, if ( !paging_mode_translate(p2m->domain) ) { - if ( !need_iommu_pt_sync(d) ) + if ( !has_iommu_pt(d) ) return 0; return iommu_legacy_map(d, _dfn(gfn_l), _mfn(gfn_l), PAGE_ORDER_4K, IOMMUF_readable | IOMMUF_writable); @@ -1432,7 +1432,7 @@ int clear_identity_p2m_entry(struct domain *d, unsigned long gfn_l) if ( !paging_mode_translate(d) ) { - if ( !need_iommu_pt_sync(d) ) + if ( !has_iommu_pt(d) ) return 0; return iommu_legacy_unmap(d, _dfn(gfn_l), PAGE_ORDER_4K); }
The current usage of need_iommu_pt_sync in p2m for non-translated guests is wrong because it doesn't correctly handle a relaxed PV hardware domain, that has need_sync set to false, but still need entries to be added from calls to {set/clear}_identity_p2m_entry. Adjust the code in guest_physmap_add_page to also check whether the domain has an iommu instead of whether it needs syncing or not in order to take a reference to a page to be mapped. Signed-off-by: Roger Pau Monné <roger.pau@citrix.com> --- Cc: George Dunlap <george.dunlap@eu.citrix.com> Cc: Jan Beulich <jbeulich@suse.com> Cc: Andrew Cooper <andrew.cooper3@citrix.com> Cc: Wei Liu <wl@xen.org> Cc: Paul Durrant <Paul.Durrant@citrix.com> --- xen/arch/x86/mm/p2m.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)