x86/sgx: Silence softlockup detection when releasing large enclaves
authorReinette Chatre <reinette.chatre@intel.com>
Tue, 8 Feb 2022 18:48:07 +0000 (10:48 -0800)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 16 Feb 2022 11:56:39 +0000 (12:56 +0100)
commit 8795359e35bc33bf86b6d0765aa7f37431db3b9c upstream.

Vijay reported that the "unclobbered_vdso_oversubscribed" selftest
triggers the softlockup detector.

Actual SGX systems have 128GB of enclave memory or more.  The
"unclobbered_vdso_oversubscribed" selftest creates one enclave which
consumes all of the enclave memory on the system. Tearing down such a
large enclave takes around a minute, most of it in the loop where
the EREMOVE instruction is applied to each individual 4k enclave page.

Spending one minute in a loop triggers the softlockup detector.

Add a cond_resched() to give other tasks a chance to run and placate
the softlockup detector.

Cc: stable@vger.kernel.org
Fixes: 1728ab54b4be ("x86/sgx: Add a page reclaimer")
Reported-by: Vijay Dhanraj <vijay.dhanraj@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
Acked-by: Dave Hansen <dave.hansen@linux.intel.com>
Tested-by: Jarkko Sakkinen <jarkko@kernel.org> (kselftest as sanity check)
Link: https://lkml.kernel.org/r/ced01cac1e75f900251b0a4ae1150aa8ebd295ec.1644345232.git.reinette.chatre@intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
arch/x86/kernel/cpu/sgx/encl.c

index 001808e..48afe96 100644 (file)
@@ -410,6 +410,8 @@ void sgx_encl_release(struct kref *ref)
                }
 
                kfree(entry);
+               /* Invoke scheduler to prevent soft lockups. */
+               cond_resched();
        }
 
        xa_destroy(&encl->page_array);