| Message ID | 20230712230202.47929-18-haitao.huang@linux.intel.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | Add Cgroup support for SGX EPC memory | expand |
On Wed Jul 12, 2023 at 11:01 PM UTC, Haitao Huang wrote: > Under heavy load, the SGX EPC reclaimers (ksgxd or future EPC cgroup > worker) may reclaim SECS EPC page for an enclave and set > encl->secs.epc_page to NULL. But the SECS EPC page is required for EAUG > in #PF handler and is used without checking for NULL and reloading. > > Fix this by checking if SECS is loaded before EAUG and load it if it was > reclaimed. > > Signed-off-by: Haitao Huang <haitao.huang@linux.intel.com> A bug fix should be 1/*. BR, Jarkko
On Mon Jul 17, 2023 at 12:48 PM UTC, Jarkko Sakkinen wrote: > On Wed Jul 12, 2023 at 11:01 PM UTC, Haitao Huang wrote: > > Under heavy load, the SGX EPC reclaimers (ksgxd or future EPC cgroup > > worker) may reclaim SECS EPC page for an enclave and set > > encl->secs.epc_page to NULL. But the SECS EPC page is required for EAUG > > in #PF handler and is used without checking for NULL and reloading. > > > > Fix this by checking if SECS is loaded before EAUG and load it if it was > > reclaimed. > > > > Signed-off-by: Haitao Huang <haitao.huang@linux.intel.com> > > A bug fix should be 1/*. And a fixes tag. Or is there a bug that is momentized by the earlier patches? This patch feels confusing to say the least. BR, Jarkko
On Mon, 17 Jul 2023 07:49:27 -0500, Jarkko Sakkinen <jarkko@kernel.org> wrote: > On Mon Jul 17, 2023 at 12:48 PM UTC, Jarkko Sakkinen wrote: >> On Wed Jul 12, 2023 at 11:01 PM UTC, Haitao Huang wrote: >> > Under heavy load, the SGX EPC reclaimers (ksgxd or future EPC cgroup >> > worker) may reclaim SECS EPC page for an enclave and set >> > encl->secs.epc_page to NULL. But the SECS EPC page is required for >> EAUG >> > in #PF handler and is used without checking for NULL and reloading. >> > >> > Fix this by checking if SECS is loaded before EAUG and load it if it >> was >> > reclaimed. >> > >> > Signed-off-by: Haitao Huang <haitao.huang@linux.intel.com> >> >> A bug fix should be 1/*. > > And a fixes tag. > > Or is there a bug that is momentized by the earlier patches? This patch > feels confusing to say the least. > It happens in heavy reclaiming cases, just extremely rare when EPC accounting is not partitioned into cgroups. Will add fix tag with the related EDMM patch. And move this as the first patch. Thanks Haitao
On Mon Jul 17, 2023 at 1:14 PM UTC, Haitao Huang wrote: > On Mon, 17 Jul 2023 07:49:27 -0500, Jarkko Sakkinen <jarkko@kernel.org> > wrote: > > > On Mon Jul 17, 2023 at 12:48 PM UTC, Jarkko Sakkinen wrote: > >> On Wed Jul 12, 2023 at 11:01 PM UTC, Haitao Huang wrote: > >> > Under heavy load, the SGX EPC reclaimers (ksgxd or future EPC cgroup > >> > worker) may reclaim SECS EPC page for an enclave and set > >> > encl->secs.epc_page to NULL. But the SECS EPC page is required for > >> EAUG > >> > in #PF handler and is used without checking for NULL and reloading. > >> > > >> > Fix this by checking if SECS is loaded before EAUG and load it if it > >> was > >> > reclaimed. > >> > > >> > Signed-off-by: Haitao Huang <haitao.huang@linux.intel.com> > >> > >> A bug fix should be 1/*. > > > > And a fixes tag. > > > > Or is there a bug that is momentized by the earlier patches? This patch > > feels confusing to say the least. > > > > It happens in heavy reclaiming cases, just extremely rare when EPC > accounting is not partitioned into cgroups. Will add fix tag with the > related EDMM patch. And move this as the first patch. I understand, it is just a good practice to follow, i.e. have prelude and then the "real" changes :-) BR, Jarkko
On 7/17/23 05:48, Jarkko Sakkinen wrote: > On Wed Jul 12, 2023 at 11:01 PM UTC, Haitao Huang wrote: >> Under heavy load, the SGX EPC reclaimers (ksgxd or future EPC cgroup >> worker) may reclaim SECS EPC page for an enclave and set >> encl->secs.epc_page to NULL. But the SECS EPC page is required for EAUG >> in #PF handler and is used without checking for NULL and reloading. >> >> Fix this by checking if SECS is loaded before EAUG and load it if it was >> reclaimed. >> >> Signed-off-by: Haitao Huang <haitao.huang@linux.intel.com> > A bug fix should be 1/*. No, bug fixes should not even be _part_ of another series. Send bug fixes separately, please.
On Mon, 17 Jul 2023 10:49:03 -0500, Dave Hansen <dave.hansen@intel.com> wrote: > On 7/17/23 05:48, Jarkko Sakkinen wrote: >> On Wed Jul 12, 2023 at 11:01 PM UTC, Haitao Huang wrote: >>> Under heavy load, the SGX EPC reclaimers (ksgxd or future EPC cgroup >>> worker) may reclaim SECS EPC page for an enclave and set >>> encl->secs.epc_page to NULL. But the SECS EPC page is required for EAUG >>> in #PF handler and is used without checking for NULL and reloading. >>> >>> Fix this by checking if SECS is loaded before EAUG and load it if it >>> was >>> reclaimed. >>> >>> Signed-off-by: Haitao Huang <haitao.huang@linux.intel.com> >> A bug fix should be 1/*. > > No, bug fixes should not even be _part_ of another series. Send bug > fixes separately, please. I sent the two bug fixes separately now. Do you want me resend this series without those? Thanks Haitao
On Mon Jul 17, 2023 at 3:49 PM UTC, Dave Hansen wrote: > On 7/17/23 05:48, Jarkko Sakkinen wrote: > > On Wed Jul 12, 2023 at 11:01 PM UTC, Haitao Huang wrote: > >> Under heavy load, the SGX EPC reclaimers (ksgxd or future EPC cgroup > >> worker) may reclaim SECS EPC page for an enclave and set > >> encl->secs.epc_page to NULL. But the SECS EPC page is required for EAUG > >> in #PF handler and is used without checking for NULL and reloading. > >> > >> Fix this by checking if SECS is loaded before EAUG and load it if it was > >> reclaimed. > >> > >> Signed-off-by: Haitao Huang <haitao.huang@linux.intel.com> > > A bug fix should be 1/*. > > No, bug fixes should not even be _part_ of another series. Send bug > fixes separately, please. Yes, that would be of course a better option. BR, Jarkko
diff --git a/arch/x86/kernel/cpu/sgx/encl.c b/arch/x86/kernel/cpu/sgx/encl.c index c321c848baa9..028d1b9d6572 100644 --- a/arch/x86/kernel/cpu/sgx/encl.c +++ b/arch/x86/kernel/cpu/sgx/encl.c @@ -235,6 +235,19 @@ static struct sgx_epc_page *sgx_encl_eldu(struct sgx_encl_page *encl_page, return epc_page; } +static struct sgx_epc_page *sgx_encl_load_secs(struct sgx_encl *encl) +{ + struct sgx_epc_page *epc_page = encl->secs.epc_page; + + if (!epc_page) { + epc_page = sgx_encl_eldu(&encl->secs, NULL); + if (!IS_ERR(epc_page)) + sgx_record_epc_page(epc_page, SGX_EPC_OWNER_ENCL_PAGE | + SGX_EPC_PAGE_UNRECLAIMABLE); + } + return epc_page; +} + static struct sgx_encl_page *__sgx_encl_load_page(struct sgx_encl *encl, struct sgx_encl_page *entry) { @@ -248,13 +261,9 @@ static struct sgx_encl_page *__sgx_encl_load_page(struct sgx_encl *encl, return entry; } - if (!(encl->secs.epc_page)) { - epc_page = sgx_encl_eldu(&encl->secs, NULL); - if (IS_ERR(epc_page)) - return ERR_CAST(epc_page); - sgx_record_epc_page(epc_page, SGX_EPC_OWNER_ENCL_PAGE | - SGX_EPC_PAGE_UNRECLAIMABLE); - } + epc_page = sgx_encl_load_secs(encl); + if (IS_ERR(epc_page)) + return ERR_CAST(epc_page); epc_page = sgx_encl_eldu(entry, encl->secs.epc_page); if (IS_ERR(epc_page)) @@ -342,6 +351,13 @@ static vm_fault_t sgx_encl_eaug_page(struct vm_area_struct *vma, mutex_lock(&encl->lock); + epc_page = sgx_encl_load_secs(encl); + if (IS_ERR(epc_page)) { + if (PTR_ERR(epc_page) == -EBUSY) + vmret = VM_FAULT_NOPAGE; + goto err_out_unlock; + } + epc_page = sgx_alloc_epc_page(encl_page, false); if (IS_ERR(epc_page)) { if (PTR_ERR(epc_page) == -EBUSY) diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c index 9ea487469e4c..68c89d575abc 100644 --- a/arch/x86/kernel/cpu/sgx/main.c +++ b/arch/x86/kernel/cpu/sgx/main.c @@ -265,6 +265,10 @@ static void sgx_reclaimer_write(struct sgx_epc_page *epc_page, mutex_lock(&encl->lock); + /* Should not be possible */ + if (WARN_ON(!(encl->secs.epc_page))) + goto out; + sgx_encl_ewb(epc_page, backing); encl_page->epc_page = NULL; encl->secs_child_cnt--;
Under heavy load, the SGX EPC reclaimers (ksgxd or future EPC cgroup worker) may reclaim SECS EPC page for an enclave and set encl->secs.epc_page to NULL. But the SECS EPC page is required for EAUG in #PF handler and is used without checking for NULL and reloading. Fix this by checking if SECS is loaded before EAUG and load it if it was reclaimed. Signed-off-by: Haitao Huang <haitao.huang@linux.intel.com> --- arch/x86/kernel/cpu/sgx/encl.c | 30 +++++++++++++++++++++++------- arch/x86/kernel/cpu/sgx/main.c | 4 ++++ 2 files changed, 27 insertions(+), 7 deletions(-)