The primary change here is patch 2, with the others being cleanup noticed to be worthwhile along the road. 1: don't ignore p2m_remove_page()'s return value 2: don't assert that the passed in MFN matches for a remove 3: make p2m_remove_page()'s parameters type-safe 4: drop pointless nested variable from guest_physmap_add_entry() 5: use available local variable in guest_physmap_add_entry() Jan
It's not very nice to return from guest_physmap_add_entry() after perhaps already having made some changes to the P2M, but this is pre- existing practice in the function, and imo better than ignoring errors. Take the liberty and replace an mfn_add() instance with a local variable already holding the result (as proven by the check immediately ahead). Signed-off-by: Jan Beulich <jbeulich@suse.com> Reviewed-by: Paul Durrant <paul.durrant@citrix.com> --- a/xen/arch/x86/mm/p2m.c +++ b/xen/arch/x86/mm/p2m.c @@ -767,8 +767,7 @@ void p2m_final_teardown(struct domain *d p2m_teardown_hostp2m(d); } - -static int +static int __must_check p2m_remove_page(struct p2m_domain *p2m, unsigned long gfn_l, unsigned long mfn, unsigned int page_order) { @@ -973,9 +972,9 @@ guest_physmap_add_entry(struct domain *d ASSERT(mfn_valid(omfn)); P2M_DEBUG("old gfn=%#lx -> mfn %#lx\n", gfn_x(ogfn) , mfn_x(omfn)); - if ( mfn_eq(omfn, mfn_add(mfn, i)) ) - p2m_remove_page(p2m, gfn_x(ogfn), mfn_x(mfn_add(mfn, i)), - 0); + if ( mfn_eq(omfn, mfn_add(mfn, i)) && + (rc = p2m_remove_page(p2m, gfn_x(ogfn), mfn_x(omfn), 0)) ) + goto out; } } } @@ -997,6 +996,7 @@ guest_physmap_add_entry(struct domain *d } } +out: p2m_unlock(p2m); return rc; @@ -2705,9 +2705,9 @@ int p2m_change_altp2m_gfn(struct domain if ( gfn_eq(new_gfn, INVALID_GFN) ) { mfn = ap2m->get_entry(ap2m, old_gfn, &t, &a, 0, NULL, NULL); - if ( mfn_valid(mfn) ) - p2m_remove_page(ap2m, gfn_x(old_gfn), mfn_x(mfn), PAGE_ORDER_4K); - rc = 0; + rc = mfn_valid(mfn) + ? p2m_remove_page(ap2m, gfn_x(old_gfn), mfn_x(mfn), PAGE_ORDER_4K) + : 0; goto out; }
On 01/04/2020 12:38, Jan Beulich wrote: > It's not very nice to return from guest_physmap_add_entry() after > perhaps already having made some changes to the P2M, but this is pre- > existing practice in the function, and imo better than ignoring errors. > > Take the liberty and replace an mfn_add() instance with a local variable > already holding the result (as proven by the check immediately ahead). > > Signed-off-by: Jan Beulich <jbeulich@suse.com> > Reviewed-by: Paul Durrant <paul.durrant@citrix.com> Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>
guest_physmap_remove_page() gets handed an MFN from the outside, yet takes the necessary lock to prevent further changes to the GFN <-> MFN mapping itself. While some callers, in particular guest_remove_page() (by way of having called get_gfn_query()), hold the GFN lock already, various others (most notably perhaps the 2nd instance in xenmem_add_to_physmap_one()) don't. While it also is an option to fix all the callers, deal with the issue in p2m_remove_page() instead: Replace the ASSERT() by a conditional and split the loop into two, such that all checking gets done before any modification would occur. Signed-off-by: Jan Beulich <jbeulich@suse.com> Reviewed-by: Paul Durrant <paul.durrant@citrix.com> --- a/xen/arch/x86/mm/p2m.c +++ b/xen/arch/x86/mm/p2m.c @@ -773,7 +773,6 @@ p2m_remove_page(struct p2m_domain *p2m, { unsigned long i; gfn_t gfn = _gfn(gfn_l); - mfn_t mfn_return; p2m_type_t t; p2m_access_t a; @@ -784,15 +783,26 @@ p2m_remove_page(struct p2m_domain *p2m, ASSERT(gfn_locked_by_me(p2m, gfn)); P2M_DEBUG("removing gfn=%#lx mfn=%#lx\n", gfn_l, mfn); + for ( i = 0; i < (1UL << page_order); ) + { + unsigned int cur_order; + mfn_t mfn_return = p2m->get_entry(p2m, gfn_add(gfn, i), &t, &a, 0, + &cur_order, NULL); + + if ( p2m_is_valid(t) && + (!mfn_valid(_mfn(mfn)) || mfn + i != mfn_x(mfn_return)) ) + return -EILSEQ; + + i += (1UL << cur_order) - ((gfn_l + i) & ((1UL << cur_order) - 1)); + } + if ( mfn_valid(_mfn(mfn)) ) { for ( i = 0; i < (1UL << page_order); i++ ) { - mfn_return = p2m->get_entry(p2m, gfn_add(gfn, i), &t, &a, 0, - NULL, NULL); + p2m->get_entry(p2m, gfn_add(gfn, i), &t, &a, 0, NULL, NULL); if ( !p2m_is_grant(t) && !p2m_is_shared(t) && !p2m_is_foreign(t) ) set_gpfn_from_mfn(mfn+i, INVALID_M2P_ENTRY); - ASSERT( !p2m_is_valid(t) || mfn + i == mfn_x(mfn_return) ); } } return p2m_set_entry(p2m, gfn, INVALID_MFN, page_order, p2m_invalid,
On 01/04/2020 12:39, Jan Beulich wrote: > guest_physmap_remove_page() gets handed an MFN from the outside, yet > takes the necessary lock to prevent further changes to the GFN <-> MFN > mapping itself. While some callers, in particular guest_remove_page() > (by way of having called get_gfn_query()), hold the GFN lock already, > various others (most notably perhaps the 2nd instance in > xenmem_add_to_physmap_one()) don't. While it also is an option to fix > all the callers, deal with the issue in p2m_remove_page() instead: > Replace the ASSERT() by a conditional and split the loop into two, such > that all checking gets done before any modification would occur. > > Signed-off-by: Jan Beulich <jbeulich@suse.com> > Reviewed-by: Paul Durrant <paul.durrant@citrix.com> Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>
Also add a couple of blank lines. Signed-off-by: Jan Beulich <jbeulich@suse.com> --- a/xen/arch/x86/mm/p2m.c +++ b/xen/arch/x86/mm/p2m.c @@ -768,11 +768,10 @@ void p2m_final_teardown(struct domain *d } static int __must_check -p2m_remove_page(struct p2m_domain *p2m, unsigned long gfn_l, unsigned long mfn, +p2m_remove_page(struct p2m_domain *p2m, gfn_t gfn, mfn_t mfn, unsigned int page_order) { unsigned long i; - gfn_t gfn = _gfn(gfn_l); p2m_type_t t; p2m_access_t a; @@ -781,7 +780,7 @@ p2m_remove_page(struct p2m_domain *p2m, return 0; ASSERT(gfn_locked_by_me(p2m, gfn)); - P2M_DEBUG("removing gfn=%#lx mfn=%#lx\n", gfn_l, mfn); + P2M_DEBUG("removing gfn=%#lx mfn=%#lx\n", gfn_x(gfn), mfn_x(mfn)); for ( i = 0; i < (1UL << page_order); ) { @@ -790,21 +789,23 @@ p2m_remove_page(struct p2m_domain *p2m, &cur_order, NULL); if ( p2m_is_valid(t) && - (!mfn_valid(_mfn(mfn)) || mfn + i != mfn_x(mfn_return)) ) + (!mfn_valid(mfn) || !mfn_eq(mfn_add(mfn, i), mfn_return)) ) return -EILSEQ; - i += (1UL << cur_order) - ((gfn_l + i) & ((1UL << cur_order) - 1)); + i += (1UL << cur_order) - + (gfn_x(gfn_add(gfn, i)) & ((1UL << cur_order) - 1)); } - if ( mfn_valid(_mfn(mfn)) ) + if ( mfn_valid(mfn) ) { for ( i = 0; i < (1UL << page_order); i++ ) { p2m->get_entry(p2m, gfn_add(gfn, i), &t, &a, 0, NULL, NULL); if ( !p2m_is_grant(t) && !p2m_is_shared(t) && !p2m_is_foreign(t) ) - set_gpfn_from_mfn(mfn+i, INVALID_M2P_ENTRY); + set_gpfn_from_mfn(mfn_x(mfn) + i, INVALID_M2P_ENTRY); } } + return p2m_set_entry(p2m, gfn, INVALID_MFN, page_order, p2m_invalid, p2m->default_access); } @@ -815,9 +816,11 @@ guest_physmap_remove_page(struct domain { struct p2m_domain *p2m = p2m_get_hostp2m(d); int rc; + gfn_lock(p2m, gfn, page_order); - rc = p2m_remove_page(p2m, gfn_x(gfn), mfn_x(mfn), page_order); + rc = p2m_remove_page(p2m, gfn, mfn, page_order); gfn_unlock(p2m, gfn, page_order); + return rc; } @@ -983,7 +986,7 @@ guest_physmap_add_entry(struct domain *d P2M_DEBUG("old gfn=%#lx -> mfn %#lx\n", gfn_x(ogfn) , mfn_x(omfn)); if ( mfn_eq(omfn, mfn_add(mfn, i)) && - (rc = p2m_remove_page(p2m, gfn_x(ogfn), mfn_x(omfn), 0)) ) + (rc = p2m_remove_page(p2m, ogfn, omfn, 0)) ) goto out; } } @@ -2716,7 +2719,7 @@ int p2m_change_altp2m_gfn(struct domain { mfn = ap2m->get_entry(ap2m, old_gfn, &t, &a, 0, NULL, NULL); rc = mfn_valid(mfn) - ? p2m_remove_page(ap2m, gfn_x(old_gfn), mfn_x(mfn), PAGE_ORDER_4K) + ? p2m_remove_page(ap2m, old_gfn, mfn, PAGE_ORDER_4K) : 0; goto out; }
On 01/04/2020 12:39, Jan Beulich wrote: > @@ -790,21 +789,23 @@ p2m_remove_page(struct p2m_domain *p2m, > &cur_order, NULL); > > if ( p2m_is_valid(t) && > - (!mfn_valid(_mfn(mfn)) || mfn + i != mfn_x(mfn_return)) ) > + (!mfn_valid(mfn) || !mfn_eq(mfn_add(mfn, i), mfn_return)) ) > return -EILSEQ; > > - i += (1UL << cur_order) - ((gfn_l + i) & ((1UL << cur_order) - 1)); > + i += (1UL << cur_order) - > + (gfn_x(gfn_add(gfn, i)) & ((1UL << cur_order) - 1)); We're gaining an number of expressions starting to look like this, but honestly, "gfn_x(gfn) + i" is equally typesafe, shorter, and easier to read IMO. ~Andrew
On 03.04.2020 00:43, Andrew Cooper wrote: > On 01/04/2020 12:39, Jan Beulich wrote: >> @@ -790,21 +789,23 @@ p2m_remove_page(struct p2m_domain *p2m, >> &cur_order, NULL); >> >> if ( p2m_is_valid(t) && >> - (!mfn_valid(_mfn(mfn)) || mfn + i != mfn_x(mfn_return)) ) >> + (!mfn_valid(mfn) || !mfn_eq(mfn_add(mfn, i), mfn_return)) ) >> return -EILSEQ; >> >> - i += (1UL << cur_order) - ((gfn_l + i) & ((1UL << cur_order) - 1)); >> + i += (1UL << cur_order) - >> + (gfn_x(gfn_add(gfn, i)) & ((1UL << cur_order) - 1)); > > We're gaining an number of expressions starting to look like this, but > honestly, "gfn_x(gfn) + i" is equally typesafe, shorter, and easier to > read IMO. Good point - in recent reviews I've commented to the same effect on patches from Julien. This patch is way too old for me to have recalled that I did like this here, too. Will switch (also elsewhere in case I find more that I introduce). Jan
On 03.04.2020 00:43, Andrew Cooper wrote: > On 01/04/2020 12:39, Jan Beulich wrote: >> @@ -790,21 +789,23 @@ p2m_remove_page(struct p2m_domain *p2m, >> &cur_order, NULL); >> >> if ( p2m_is_valid(t) && >> - (!mfn_valid(_mfn(mfn)) || mfn + i != mfn_x(mfn_return)) ) >> + (!mfn_valid(mfn) || !mfn_eq(mfn_add(mfn, i), mfn_return)) ) >> return -EILSEQ; >> >> - i += (1UL << cur_order) - ((gfn_l + i) & ((1UL << cur_order) - 1)); >> + i += (1UL << cur_order) - >> + (gfn_x(gfn_add(gfn, i)) & ((1UL << cur_order) - 1)); > > We're gaining an number of expressions starting to look like this, but > honestly, "gfn_x(gfn) + i" is equally typesafe, shorter, and easier to > read IMO. May I, just like you said for patch 3, imply A-b with this adjusted? Jan
On 03/04/2020 10:14, Jan Beulich wrote: > On 03.04.2020 00:43, Andrew Cooper wrote: >> On 01/04/2020 12:39, Jan Beulich wrote: >>> @@ -790,21 +789,23 @@ p2m_remove_page(struct p2m_domain *p2m, >>> &cur_order, NULL); >>> >>> if ( p2m_is_valid(t) && >>> - (!mfn_valid(_mfn(mfn)) || mfn + i != mfn_x(mfn_return)) ) >>> + (!mfn_valid(mfn) || !mfn_eq(mfn_add(mfn, i), mfn_return)) ) >>> return -EILSEQ; >>> >>> - i += (1UL << cur_order) - ((gfn_l + i) & ((1UL << cur_order) - 1)); >>> + i += (1UL << cur_order) - >>> + (gfn_x(gfn_add(gfn, i)) & ((1UL << cur_order) - 1)); >> We're gaining an number of expressions starting to look like this, but >> honestly, "gfn_x(gfn) + i" is equally typesafe, shorter, and easier to >> read IMO. > May I, just like you said for patch 3, imply A-b with this adjusted? Yes. Sorry - it was late when I was reviewing. Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>
There's an outer scope rc already, and its use for the mem-sharing logic does not conflict with its use elsewhere in the function. Signed-off-by: Jan Beulich <jbeulich@suse.com> --- a/xen/arch/x86/mm/p2m.c +++ b/xen/arch/x86/mm/p2m.c @@ -905,7 +905,6 @@ guest_physmap_add_entry(struct domain *d if ( p2m_is_shared(ot) ) { /* Do an unshare to cleanly take care of all corner cases. */ - int rc; rc = mem_sharing_unshare_page(p2m->domain, gfn_x(gfn_add(gfn, i))); if ( rc ) {
On 01/04/2020 12:40, Jan Beulich wrote: > There's an outer scope rc already, and its use for the mem-sharing logic > does not conflict with its use elsewhere in the function. > > Signed-off-by: Jan Beulich <jbeulich@suse.com> Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>
The domain is being passed in - no need to obtain it from p2m->domain. Also drop a pointless cast while touching this code anyway. Signed-off-by: Jan Beulich <jbeulich@suse.com> --- a/xen/arch/x86/mm/p2m.c +++ b/xen/arch/x86/mm/p2m.c @@ -905,7 +905,7 @@ guest_physmap_add_entry(struct domain *d if ( p2m_is_shared(ot) ) { /* Do an unshare to cleanly take care of all corner cases. */ - rc = mem_sharing_unshare_page(p2m->domain, gfn_x(gfn_add(gfn, i))); + rc = mem_sharing_unshare_page(d, gfn_x(gfn_add(gfn, i))); if ( rc ) { p2m_unlock(p2m); @@ -922,8 +922,7 @@ guest_physmap_add_entry(struct domain *d * Foreign domains are okay to place an event as they * won't go to sleep. */ - (void)mem_sharing_notify_enomem(p2m->domain, - gfn_x(gfn_add(gfn, i)), false); + mem_sharing_notify_enomem(d, gfn_x(gfn_add(gfn, i)), false); return rc; } omfn = p2m->get_entry(p2m, gfn_add(gfn, i),
On 01/04/2020 12:40, Jan Beulich wrote: > The domain is being passed in - no need to obtain it from p2m->domain. > Also drop a pointless cast while touching this code anyway. > > Signed-off-by: Jan Beulich <jbeulich@suse.com> > > --- a/xen/arch/x86/mm/p2m.c > +++ b/xen/arch/x86/mm/p2m.c > @@ -905,7 +905,7 @@ guest_physmap_add_entry(struct domain *d > if ( p2m_is_shared(ot) ) > { > /* Do an unshare to cleanly take care of all corner cases. */ > - rc = mem_sharing_unshare_page(p2m->domain, gfn_x(gfn_add(gfn, i))); > + rc = mem_sharing_unshare_page(d, gfn_x(gfn_add(gfn, i))); Same as patch 3. I'd recommend "gfn_x(gfn) + i" in preference (seeing as you're cleaning up this line anyway). Acked-by: Andrew Cooper <andrew.cooper3@citrix.com> seeing as you didn't introduce it, but preferably with it changed. ~Andrew
© 2016 - 2024 Red Hat, Inc.