* [PATCH v2] Fix the issue further discussed in:
@ 2020-10-05 3:17 Jarkko Sakkinen
2020-10-05 11:11 ` Matthew Wilcox
0 siblings, 1 reply; 4+ messages in thread
From: Jarkko Sakkinen @ 2020-10-05 3:17 UTC (permalink / raw)
To: linux-sgx
Cc: Jarkko Sakkinen, Haitao Huang, Matthew Wilcox,
Sean Christopherson, Jethro Beekman, Dave Hansen
1. https://lore.kernel.org/linux-sgx/op.0rwbv916wjvjmi@mqcpg7oapc828.gar.corp.intel.com/
2. https://lore.kernel.org/linux-sgx/20201003195440.GD20115@casper.infradead.org/
Use the approach suggested by Matthew, and supported by the analysis
that I wrote:
https://lore.kernel.org/linux-sgx/20201005030619.GA126283@linux.intel.com/
Reported-by: Haitao Huang <haitao.huang@linux.intel.com>
Suggested-by: Matthew Wilcox <willy@infradead.org>
Cc: Sean Christopherson <sean.j.christopherson@intel.com>
Cc: Jethro Beekman <jethro@fortanix.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>
---
arch/x86/kernel/cpu/sgx/encl.c | 24 +++++++++++++++++++++++-
1 file changed, 23 insertions(+), 1 deletion(-)
diff --git a/arch/x86/kernel/cpu/sgx/encl.c b/arch/x86/kernel/cpu/sgx/encl.c
index 4c6407cd857a..2bb3ec6996e9 100644
--- a/arch/x86/kernel/cpu/sgx/encl.c
+++ b/arch/x86/kernel/cpu/sgx/encl.c
@@ -307,6 +307,7 @@ int sgx_encl_may_map(struct sgx_encl *encl, unsigned long start,
unsigned long idx_start = PFN_DOWN(start);
unsigned long idx_end = PFN_DOWN(end - 1);
struct sgx_encl_page *page;
+ unsigned long count = 0;
XA_STATE(xas, &encl->page_array, idx_start);
@@ -317,10 +318,31 @@ int sgx_encl_may_map(struct sgx_encl *encl, unsigned long start,
if (current->personality & READ_IMPLIES_EXEC)
return -EACCES;
- xas_for_each(&xas, page, idx_end)
+ /*
+ * No need to hold encl->lock:
+ * 1. None of the page->* get written.
+ * 2. page->vm_max_prot_bits is set in sgx_encl_page_alloc(). This
+ * is before calling xa_insert(). After that it is never modified.
+ */
+ xas_lock(&xas);
+ xas_for_each(&xas, page, idx_end) {
+ if (++count % XA_CHECK_SCHED)
+ continue;
+
+ xas_pause(&xas);
+ xas_unlock(&xas);
+
+ /*
+ * Attributes are not protected by the xa_lock, so I'm assuming
+ * that this is the legit place for the check.
+ */
if (!page || (~page->vm_max_prot_bits & vm_prot_bits))
return -EACCES;
+ cond_resched();
+ xas_lock(&xas);
+ }
+
return 0;
}
--
2.25.1
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH v2] Fix the issue further discussed in:
2020-10-05 3:17 [PATCH v2] Fix the issue further discussed in: Jarkko Sakkinen
@ 2020-10-05 11:11 ` Matthew Wilcox
2020-10-05 11:48 ` Jarkko Sakkinen
0 siblings, 1 reply; 4+ messages in thread
From: Matthew Wilcox @ 2020-10-05 11:11 UTC (permalink / raw)
To: Jarkko Sakkinen
Cc: linux-sgx, Haitao Huang, Sean Christopherson, Jethro Beekman,
Dave Hansen
On Mon, Oct 05, 2020 at 06:17:59AM +0300, Jarkko Sakkinen wrote:
> @@ -317,10 +318,31 @@ int sgx_encl_may_map(struct sgx_encl *encl, unsigned long start,
> if (current->personality & READ_IMPLIES_EXEC)
> return -EACCES;
>
> - xas_for_each(&xas, page, idx_end)
> + /*
> + * No need to hold encl->lock:
> + * 1. None of the page->* get written.
> + * 2. page->vm_max_prot_bits is set in sgx_encl_page_alloc(). This
> + * is before calling xa_insert(). After that it is never modified.
> + */
> + xas_lock(&xas);
> + xas_for_each(&xas, page, idx_end) {
> + if (++count % XA_CHECK_SCHED)
> + continue;
This really doesn't do what you think it does.
int ret = 0;
int count = 0;
xas_lock(&xas);
while (xas.index < idx_end) {
struct sgx_page *page = xas_next(&xas);
if (!page || (~page->vm_max_prot_bits & vm_prot_bits)) {
ret = -EACCESS;
break;
}
if (++count % XA_CHECK_SCHED)
continue;
xas_pause(&xas);
xas_unlock(&xas);
cond_resched();
xas_lock(&xas);
}
xas_unlock(&xas);
return ret;
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH v2] Fix the issue further discussed in:
2020-10-05 11:11 ` Matthew Wilcox
@ 2020-10-05 11:48 ` Jarkko Sakkinen
2020-10-05 11:48 ` Jarkko Sakkinen
0 siblings, 1 reply; 4+ messages in thread
From: Jarkko Sakkinen @ 2020-10-05 11:48 UTC (permalink / raw)
To: Matthew Wilcox
Cc: linux-sgx, Haitao Huang, Sean Christopherson, Jethro Beekman,
Dave Hansen
On Mon, Oct 05, 2020 at 12:11:39PM +0100, Matthew Wilcox wrote:
> On Mon, Oct 05, 2020 at 06:17:59AM +0300, Jarkko Sakkinen wrote:
> > @@ -317,10 +318,31 @@ int sgx_encl_may_map(struct sgx_encl *encl, unsigned long start,
> > if (current->personality & READ_IMPLIES_EXEC)
> > return -EACCES;
> >
> > - xas_for_each(&xas, page, idx_end)
> > + /*
> > + * No need to hold encl->lock:
> > + * 1. None of the page->* get written.
> > + * 2. page->vm_max_prot_bits is set in sgx_encl_page_alloc(). This
> > + * is before calling xa_insert(). After that it is never modified.
> > + */
> > + xas_lock(&xas);
> > + xas_for_each(&xas, page, idx_end) {
> > + if (++count % XA_CHECK_SCHED)
> > + continue;
>
> This really doesn't do what you think it does.
>
> int ret = 0;
> int count = 0;
>
> xas_lock(&xas);
> while (xas.index < idx_end) {
> struct sgx_page *page = xas_next(&xas);
>
> if (!page || (~page->vm_max_prot_bits & vm_prot_bits)) {
> ret = -EACCESS;
> break;
> }
>
> if (++count % XA_CHECK_SCHED)
> continue;
> xas_pause(&xas);
> xas_unlock(&xas);
> cond_resched();
> xas_lock(&xas);
> }
> xas_unlock(&xas);
>
> return ret;
No mine certainly does not, it locks up the system if the loop succeeds
(i.e. does not return -EACCESS) :-) Unfortunately had by mistake the v1
patch (xa_load()) in the kernel that I used to test.
/Jarkko
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH v2] Fix the issue further discussed in:
2020-10-05 11:48 ` Jarkko Sakkinen
@ 2020-10-05 11:48 ` Jarkko Sakkinen
0 siblings, 0 replies; 4+ messages in thread
From: Jarkko Sakkinen @ 2020-10-05 11:48 UTC (permalink / raw)
To: Matthew Wilcox
Cc: linux-sgx, Haitao Huang, Sean Christopherson, Jethro Beekman,
Dave Hansen
On Mon, Oct 05, 2020 at 02:48:07PM +0300, Jarkko Sakkinen wrote:
> On Mon, Oct 05, 2020 at 12:11:39PM +0100, Matthew Wilcox wrote:
> > On Mon, Oct 05, 2020 at 06:17:59AM +0300, Jarkko Sakkinen wrote:
> > > @@ -317,10 +318,31 @@ int sgx_encl_may_map(struct sgx_encl *encl, unsigned long start,
> > > if (current->personality & READ_IMPLIES_EXEC)
> > > return -EACCES;
> > >
> > > - xas_for_each(&xas, page, idx_end)
> > > + /*
> > > + * No need to hold encl->lock:
> > > + * 1. None of the page->* get written.
> > > + * 2. page->vm_max_prot_bits is set in sgx_encl_page_alloc(). This
> > > + * is before calling xa_insert(). After that it is never modified.
> > > + */
> > > + xas_lock(&xas);
> > > + xas_for_each(&xas, page, idx_end) {
> > > + if (++count % XA_CHECK_SCHED)
> > > + continue;
> >
> > This really doesn't do what you think it does.
> >
> > int ret = 0;
> > int count = 0;
> >
> > xas_lock(&xas);
> > while (xas.index < idx_end) {
> > struct sgx_page *page = xas_next(&xas);
> >
> > if (!page || (~page->vm_max_prot_bits & vm_prot_bits)) {
> > ret = -EACCESS;
> > break;
> > }
> >
> > if (++count % XA_CHECK_SCHED)
> > continue;
> > xas_pause(&xas);
> > xas_unlock(&xas);
> > cond_resched();
> > xas_lock(&xas);
> > }
> > xas_unlock(&xas);
> >
> > return ret;
>
> No mine certainly does not, it locks up the system if the loop succeeds
> (i.e. does not return -EACCESS) :-) Unfortunately had by mistake the v1
> patch (xa_load()) in the kernel that I used to test.
... and not having xas_unlock() in the end was not intentional.
/Jarkko
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2020-10-05 11:49 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-10-05 3:17 [PATCH v2] Fix the issue further discussed in: Jarkko Sakkinen
2020-10-05 11:11 ` Matthew Wilcox
2020-10-05 11:48 ` Jarkko Sakkinen
2020-10-05 11:48 ` Jarkko Sakkinen
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).