| Message ID | 20210809093127.76264-2-jarkko@kernel.org (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | Add an oversubscription test | expand |
On 8/9/21 3:31 AM, Jarkko Sakkinen wrote: > Just like normal memory, SGX memory can be overcommitted. SGX has its > own reclaim mechanism which kicks in when physical SGX memory (Enclave > Page Cache / EPC) is exhausted. That reclaim mechanism is relatively > rarely exercised and needs selftests to poke at it. > > The amount of EPC on the system is determined by the BIOS and it varies > wildly between systems. It can be dozens of MB on desktops, or many GB > on servers. > > To run in a reasonable amount of time, the selftest needs to know how > much EPC there is in the system. > > Introduce a new debugfs file to export that information. > > Acked-by: Dave Hansen <dave.hansen@linux.intel.com> > Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org> > --- > > v3: > * Describe the units of sgx_total_mem in Dcumentation/x86/sgx.rst. > * Rewrite of the commit message (suggested by Dave): > https://lore.kernel.org/linux-sgx/5d3614af-2393-6744-9d85-7001241ca76e@intel.com/ > > v2: > * sgx_nr_all_pages -> sgx_total_mem > > Documentation/x86/sgx.rst | 6 ++++++ > arch/x86/kernel/cpu/sgx/main.c | 10 +++++++++- > 2 files changed, 15 insertions(+), 1 deletion(-) > > diff --git a/Documentation/x86/sgx.rst b/Documentation/x86/sgx.rst > index dd0ac96ff9ef..f11bfb331b93 100644 > --- a/Documentation/x86/sgx.rst > +++ b/Documentation/x86/sgx.rst > @@ -250,3 +250,9 @@ user wants to deploy SGX applications both on the host and in guests > on the same machine, the user should reserve enough EPC (by taking out > total virtual EPC size of all SGX VMs from the physical EPC size) for > host SGX applications so they can run with acceptable performance. > + > +Debugging > +========= > + > +*/sys/kernel/debug/x86/sgx_total_mem* contains an integer describing > +the total SGX reserved memory in bytes, available in the system. Why not add the details you have in the commit log to the document? Also add more details on how to debug/test. > diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c > index 63d3de02bbcc..b65da19a53ee 100644 > --- a/arch/x86/kernel/cpu/sgx/main.c > +++ b/arch/x86/kernel/cpu/sgx/main.c > @@ -1,6 +1,7 @@ > // SPDX-License-Identifier: GPL-2.0 > /* Copyright(c) 2016-20 Intel Corporation. */ > > +#include <linux/debugfs.h> > #include <linux/file.h> > #include <linux/freezer.h> > #include <linux/highmem.h> > @@ -28,7 +29,10 @@ static DECLARE_WAIT_QUEUE_HEAD(ksgxd_waitq); > static LIST_HEAD(sgx_active_page_list); > static DEFINE_SPINLOCK(sgx_reclaimer_lock); > > -/* The free page list lock protected variables prepend the lock. */ > +/* Total EPC memory available in bytes. */ > +static unsigned long sgx_total_mem; > + > +/* The number of free EPC pages in all nodes. */ > static unsigned long sgx_nr_free_pages; > > /* Nodes with one or more EPC sections. */ > @@ -656,6 +660,8 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size, > list_add_tail(§ion->pages[i].list, &sgx_dirty_page_list); > } > > + sgx_total_mem += nr_pages * PAGE_SIZE; > + > return true; > } > > @@ -823,6 +829,8 @@ static int __init sgx_init(void) > if (sgx_vepc_init() && ret) > goto err_provision; > > + debugfs_create_ulong("sgx_total_mem", 0444, arch_debugfs_dir, &sgx_total_mem); > + > return 0; > > err_provision: > thanks, -- Shuah
On 8/9/21 3:31 AM, Jarkko Sakkinen wrote: > Just like normal memory, SGX memory can be overcommitted. SGX has its > own reclaim mechanism which kicks in when physical SGX memory (Enclave > Page Cache / EPC) is exhausted. That reclaim mechanism is relatively > rarely exercised and needs selftests to poke at it. > > The amount of EPC on the system is determined by the BIOS and it varies > wildly between systems. It can be dozens of MB on desktops, or many GB > on servers. > > To run in a reasonable amount of time, the selftest needs to know how > much EPC there is in the system. > > Introduce a new debugfs file to export that information. > > Acked-by: Dave Hansen <dave.hansen@linux.intel.com> > Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org> > --- > > v3: > * Describe the units of sgx_total_mem in Dcumentation/x86/sgx.rst. > * Rewrite of the commit message (suggested by Dave): > https://lore.kernel.org/linux-sgx/5d3614af-2393-6744-9d85-7001241ca76e@intel.com/ > > v2: > * sgx_nr_all_pages -> sgx_total_mem > > Documentation/x86/sgx.rst | 6 ++++++ > arch/x86/kernel/cpu/sgx/main.c | 10 +++++++++- > 2 files changed, 15 insertions(+), 1 deletion(-) > > diff --git a/Documentation/x86/sgx.rst b/Documentation/x86/sgx.rst > index dd0ac96ff9ef..f11bfb331b93 100644 > --- a/Documentation/x86/sgx.rst > +++ b/Documentation/x86/sgx.rst > @@ -250,3 +250,9 @@ user wants to deploy SGX applications both on the host and in guests > on the same machine, the user should reserve enough EPC (by taking out > total virtual EPC size of all SGX VMs from the physical EPC size) for > host SGX applications so they can run with acceptable performance. > + > +Debugging > +========= > + > +*/sys/kernel/debug/x86/sgx_total_mem* contains an integer describing > +the total SGX reserved memory in bytes, available in the system. > diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c > index 63d3de02bbcc..b65da19a53ee 100644 > --- a/arch/x86/kernel/cpu/sgx/main.c > +++ b/arch/x86/kernel/cpu/sgx/main.c > @@ -1,6 +1,7 @@ > // SPDX-License-Identifier: GPL-2.0 > /* Copyright(c) 2016-20 Intel Corporation. */ > > +#include <linux/debugfs.h> > #include <linux/file.h> > #include <linux/freezer.h> > #include <linux/highmem.h> > @@ -28,7 +29,10 @@ static DECLARE_WAIT_QUEUE_HEAD(ksgxd_waitq); > static LIST_HEAD(sgx_active_page_list); > static DEFINE_SPINLOCK(sgx_reclaimer_lock); > > -/* The free page list lock protected variables prepend the lock. */ > +/* Total EPC memory available in bytes. */ > +static unsigned long sgx_total_mem; > + > +/* The number of free EPC pages in all nodes. */ > static unsigned long sgx_nr_free_pages; > > /* Nodes with one or more EPC sections. */ > @@ -656,6 +660,8 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size, > list_add_tail(§ion->pages[i].list, &sgx_dirty_page_list); > } > > + sgx_total_mem += nr_pages * PAGE_SIZE; > + > return true; > } > > @@ -823,6 +829,8 @@ static int __init sgx_init(void) > if (sgx_vepc_init() && ret) > goto err_provision; > > + debugfs_create_ulong("sgx_total_mem", 0444, arch_debugfs_dir, &sgx_total_mem); dbugfs init dependency would be met I assume by this time? > + > return 0; > > err_provision: >
On Fri, 2021-08-13 at 11:33 -0600, Shuah Khan wrote: > Why not add the details you have in the commit log to the document? > Also add more details on how to debug/test. Agreed, we can re-use it, thanks for the remark. I'll extend it, but it remains to be seen where exactly it will be documented. I've started to manage this patch as a separate cycle: https://lore.kernel.org/linux-sgx/20210812224645.90280-1-jarkko@kernel.org/T/#u I'll CC you the 1st non-RFC version, and hold on with the patch set under discussion up until we get the kernel change first to the mainline. Should happen within couple of days as I think we got quite well sorted out with Dave last week, what to do next. /Jarkko
diff --git a/Documentation/x86/sgx.rst b/Documentation/x86/sgx.rst index dd0ac96ff9ef..f11bfb331b93 100644 --- a/Documentation/x86/sgx.rst +++ b/Documentation/x86/sgx.rst @@ -250,3 +250,9 @@ user wants to deploy SGX applications both on the host and in guests on the same machine, the user should reserve enough EPC (by taking out total virtual EPC size of all SGX VMs from the physical EPC size) for host SGX applications so they can run with acceptable performance. + +Debugging +========= + +*/sys/kernel/debug/x86/sgx_total_mem* contains an integer describing +the total SGX reserved memory in bytes, available in the system. diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c index 63d3de02bbcc..b65da19a53ee 100644 --- a/arch/x86/kernel/cpu/sgx/main.c +++ b/arch/x86/kernel/cpu/sgx/main.c @@ -1,6 +1,7 @@ // SPDX-License-Identifier: GPL-2.0 /* Copyright(c) 2016-20 Intel Corporation. */ +#include <linux/debugfs.h> #include <linux/file.h> #include <linux/freezer.h> #include <linux/highmem.h> @@ -28,7 +29,10 @@ static DECLARE_WAIT_QUEUE_HEAD(ksgxd_waitq); static LIST_HEAD(sgx_active_page_list); static DEFINE_SPINLOCK(sgx_reclaimer_lock); -/* The free page list lock protected variables prepend the lock. */ +/* Total EPC memory available in bytes. */ +static unsigned long sgx_total_mem; + +/* The number of free EPC pages in all nodes. */ static unsigned long sgx_nr_free_pages; /* Nodes with one or more EPC sections. */ @@ -656,6 +660,8 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size, list_add_tail(§ion->pages[i].list, &sgx_dirty_page_list); } + sgx_total_mem += nr_pages * PAGE_SIZE; + return true; } @@ -823,6 +829,8 @@ static int __init sgx_init(void) if (sgx_vepc_init() && ret) goto err_provision; + debugfs_create_ulong("sgx_total_mem", 0444, arch_debugfs_dir, &sgx_total_mem); + return 0; err_provision: