| Message ID | 20211029121857.477885-2-jarkko@kernel.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [v9,1/2] x86/sgx: Rename fallback labels in sgx_init() | expand |
On Fri, Oct 29, 2021 at 03:18:57PM +0300, Jarkko Sakkinen wrote: > The amount of SGX memory on the system is determined by the BIOS and it > varies wildly between systems. It can be from dozens of MB's on desktops > or VM's, up to many GB's on servers. Just like for regular memory, it is > sometimes useful to know the amount of usable SGX memory in the system. > > Add an attribute for the amount of SGX memory in bytes to each NUMA > node. The path is /sys/devices/system/node/node[0-9]*/sgx/size. > Calculate these values by summing up EPC section sizes for each node > during the driver initalization. > > Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> > Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org> > --- > v9: > * Fix racy initialization of sysfs attributes: > https://lore.kernel.org/linux-sgx/YXOsx8SvFJV5R7lU@kroah.com/ > > v8: > * Fix a bug in sgx_numa_init(): node->dev should be only set after > sysfe_create_group(). Otherwise, sysfs_remove_group() will issue a > warning in sgx_numa_exit(), when sgx_create_group() is unsuccessful, > because the group does not exist. > > v7: > * Shorten memory_size to size. The prefix makes the name only longer > but does not clarify things more than "size" would. > * Use device_attribute instead of kobj_attribute. > * Use named attribute group instead of creating raw kobject just for > the "sgx" subdirectory. > > v6: > * Initialize node->size to zero in sgx_setup_epc_section(), when the > node is first accessed. > > v5 > * A new patch based on the discussion on > https://lore.kernel.org/linux-sgx/3a7cab4115b4f902f3509ad8652e616b91703e1d.camel@kernel.org/T/#t > --- > Documentation/ABI/stable/sysfs-devices-node | 7 +++++ > arch/x86/include/asm/sgx.h | 2 ++ > arch/x86/kernel/cpu/sgx/main.c | 31 +++++++++++++++++++++ > arch/x86/kernel/cpu/sgx/sgx.h | 2 ++ > drivers/base/node.c | 16 ++++++++++- > 5 files changed, 57 insertions(+), 1 deletion(-) > > diff --git a/Documentation/ABI/stable/sysfs-devices-node b/Documentation/ABI/stable/sysfs-devices-node > index 484fc04bcc25..12dc2149e8e0 100644 > --- a/Documentation/ABI/stable/sysfs-devices-node > +++ b/Documentation/ABI/stable/sysfs-devices-node > @@ -176,3 +176,10 @@ Contact: Keith Busch <keith.busch@intel.com> > Description: > The cache write policy: 0 for write-back, 1 for write-through, > other or unknown. > + > +What: /sys/devices/system/node/nodeX/sgx/size > +Date: October 2021 > +Contact: Jarkko Sakkinen <jarkko@kernel.org> > +Description: > + Total available physical SGX memory, also known as Enclave Page > + Cache (EPC), in bytes. > diff --git a/arch/x86/include/asm/sgx.h b/arch/x86/include/asm/sgx.h > index 05f3e21f01a7..43d1ccae6c62 100644 > --- a/arch/x86/include/asm/sgx.h > +++ b/arch/x86/include/asm/sgx.h > @@ -365,6 +365,8 @@ struct sgx_sigstruct { > * comment! > */ > > +extern const struct attribute_group sgx_node_group; > + > #ifdef CONFIG_X86_SGX_KVM > int sgx_virt_ecreate(struct sgx_pageinfo *pageinfo, void __user *secs, > int *trapnr); > diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c > index a6e313f1a82d..86d6d309ff29 100644 > --- a/arch/x86/kernel/cpu/sgx/main.c > +++ b/arch/x86/kernel/cpu/sgx/main.c > @@ -714,9 +714,12 @@ static bool __init sgx_page_cache_init(void) > spin_lock_init(&sgx_numa_nodes[nid].lock); > INIT_LIST_HEAD(&sgx_numa_nodes[nid].free_page_list); > node_set(nid, sgx_numa_mask); > + sgx_numa_nodes[nid].size = 0; > } > > sgx_epc_sections[i].node = &sgx_numa_nodes[nid]; > + sgx_numa_nodes[nid].dev = &node_devices[nid]->dev; > + sgx_numa_nodes[nid].size += size; > > sgx_nr_epc_sections++; > } > @@ -790,6 +793,34 @@ int sgx_set_attribute(unsigned long *allowed_attributes, > } > EXPORT_SYMBOL_GPL(sgx_set_attribute); > > +#ifdef CONFIG_NUMA > +static ssize_t size_show(struct device *dev, struct device_attribute *attr, char *buf) > +{ > + unsigned long size = 0; > + int nid; > + > + for (nid = 0; nid < num_possible_nodes(); nid++) { > + if (dev == sgx_numa_nodes[nid].dev) { > + size = sgx_numa_nodes[nid].size; > + break; > + } > + } > + > + return sysfs_emit(buf, "%lu\n", size); > +} > +DEVICE_ATTR_RO(size); static?
On 10/29/21 5:18 AM, Jarkko Sakkinen wrote: > +What: /sys/devices/system/node/nodeX/sgx/size > +Date: October 2021 > +Contact: Jarkko Sakkinen <jarkko@kernel.org> I don't think we should do something *entirely* SGX-specific here. The only question to me is whether any non-SGX users want something like this and who they are. Here are some ideas I like more than an "sgx/" directory: /sys/devices/system/node/nodeX/arch/sgx_size /sys/devices/system/node/nodeX/x86/sgx_size /sys/devices/system/node/nodeX/coco/sgx_size There's somebody else *today* who is trying to do something in the same general area: per-node platform-specific memory encryption capabilities: > https://lore.kernel.org/all/20211027195511.207552-6-martin.fernandez@eclypsium.com/ Also, could we please think through how this will look if we add more attributes? I can imagine wanting both: * total SGX memory available * total SGX memory present But those would be quite hard to differentiate if we have just an "sgx_size". Wouldn't it be much nicer to name them things like: sgx_present_bytes sgx_available_bytes ? In other words, can we please try to think just a bit into the future on this one? What other SGX things will we want to export like this? Outside of SGX, who else wants stuff _like_ this?
On Fri, 2021-10-29 at 16:53 +0200, Greg Kroah-Hartman wrote: > On Fri, Oct 29, 2021 at 03:18:57PM +0300, Jarkko Sakkinen wrote: > > The amount of SGX memory on the system is determined by the BIOS and it > > varies wildly between systems. It can be from dozens of MB's on desktops > > or VM's, up to many GB's on servers. Just like for regular memory, it is > > sometimes useful to know the amount of usable SGX memory in the system. > > > > Add an attribute for the amount of SGX memory in bytes to each NUMA > > node. The path is /sys/devices/system/node/node[0-9]*/sgx/size. > > Calculate these values by summing up EPC section sizes for each node > > during the driver initalization. > > > > Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> > > Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org> > > --- > > v9: > > * Fix racy initialization of sysfs attributes: > > https://lore.kernel.org/linux-sgx/YXOsx8SvFJV5R7lU@kroah.com/ > > > > v8: > > * Fix a bug in sgx_numa_init(): node->dev should be only set after > > sysfe_create_group(). Otherwise, sysfs_remove_group() will issue a > > warning in sgx_numa_exit(), when sgx_create_group() is unsuccessful, > > because the group does not exist. > > > > v7: > > * Shorten memory_size to size. The prefix makes the name only longer > > but does not clarify things more than "size" would. > > * Use device_attribute instead of kobj_attribute. > > * Use named attribute group instead of creating raw kobject just for > > the "sgx" subdirectory. > > > > v6: > > * Initialize node->size to zero in sgx_setup_epc_section(), when the > > node is first accessed. > > > > v5 > > * A new patch based on the discussion on > > https://lore.kernel.org/linux-sgx/3a7cab4115b4f902f3509ad8652e616b91703e1d.camel@kernel.org/T/#t > > --- > > Documentation/ABI/stable/sysfs-devices-node | 7 +++++ > > arch/x86/include/asm/sgx.h | 2 ++ > > arch/x86/kernel/cpu/sgx/main.c | 31 +++++++++++++++++++++ > > arch/x86/kernel/cpu/sgx/sgx.h | 2 ++ > > drivers/base/node.c | 16 ++++++++++- > > 5 files changed, 57 insertions(+), 1 deletion(-) > > > > diff --git a/Documentation/ABI/stable/sysfs-devices-node b/Documentation/ABI/stable/sysfs-devices-node > > index 484fc04bcc25..12dc2149e8e0 100644 > > --- a/Documentation/ABI/stable/sysfs-devices-node > > +++ b/Documentation/ABI/stable/sysfs-devices-node > > @@ -176,3 +176,10 @@ Contact: Keith Busch <keith.busch@intel.com> > > Description: > > The cache write policy: 0 for write-back, 1 for write-through, > > other or unknown. > > + > > +What: /sys/devices/system/node/nodeX/sgx/size > > +Date: October 2021 > > +Contact: Jarkko Sakkinen <jarkko@kernel.org> > > +Description: > > + Total available physical SGX memory, also known as Enclave Page > > + Cache (EPC), in bytes. > > diff --git a/arch/x86/include/asm/sgx.h b/arch/x86/include/asm/sgx.h > > index 05f3e21f01a7..43d1ccae6c62 100644 > > --- a/arch/x86/include/asm/sgx.h > > +++ b/arch/x86/include/asm/sgx.h > > @@ -365,6 +365,8 @@ struct sgx_sigstruct { > > * comment! > > */ > > > > +extern const struct attribute_group sgx_node_group; > > + > > #ifdef CONFIG_X86_SGX_KVM > > int sgx_virt_ecreate(struct sgx_pageinfo *pageinfo, void __user *secs, > > int *trapnr); > > diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c > > index a6e313f1a82d..86d6d309ff29 100644 > > --- a/arch/x86/kernel/cpu/sgx/main.c > > +++ b/arch/x86/kernel/cpu/sgx/main.c > > @@ -714,9 +714,12 @@ static bool __init sgx_page_cache_init(void) > > spin_lock_init(&sgx_numa_nodes[nid].lock); > > INIT_LIST_HEAD(&sgx_numa_nodes[nid].free_page_list); > > node_set(nid, sgx_numa_mask); > > + sgx_numa_nodes[nid].size = 0; > > } > > > > sgx_epc_sections[i].node = &sgx_numa_nodes[nid]; > > + sgx_numa_nodes[nid].dev = &node_devices[nid]->dev; > > + sgx_numa_nodes[nid].size += size; > > > > sgx_nr_epc_sections++; > > } > > @@ -790,6 +793,34 @@ int sgx_set_attribute(unsigned long *allowed_attributes, > > } > > EXPORT_SYMBOL_GPL(sgx_set_attribute); > > > > +#ifdef CONFIG_NUMA > > +static ssize_t size_show(struct device *dev, struct device_attribute *attr, char *buf) > > +{ > > + unsigned long size = 0; > > + int nid; > > + > > + for (nid = 0; nid < num_possible_nodes(); nid++) { > > + if (dev == sgx_numa_nodes[nid].dev) { > > + size = sgx_numa_nodes[nid].size; > > + break; > > + } > > + } > > + > > + return sysfs_emit(buf, "%lu\n", size); > > +} > > +DEVICE_ATTR_RO(size); > > static? The named ("sgx") attribute group is exported: extern const struct attribute_group sgx_node_group; and added as an additional group of the node device: static const struct attribute_group node_dev_group = { .attrs = node_dev_attrs, }; static const struct attribute_group *node_dev_groups[] = { &node_dev_group, #ifdef CONFIG_X86_SGX &sgx_node_group, #endif /* CONFIG_X86_SGX */ NULL, }; An alternative would to define a helper, e.g. ssize_t sgx_get_node_size(struct device *dev); And use this in drivers/base/node.c to implement the attribute. I'm fine either way... /Jarkko
On Fri, 2021-10-29 at 10:00 -0700, Dave Hansen wrote: > On 10/29/21 5:18 AM, Jarkko Sakkinen wrote: > > +What: /sys/devices/system/node/nodeX/sgx/size > > +Date: October 2021 > > +Contact: Jarkko Sakkinen <jarkko@kernel.org> > > I don't think we should do something *entirely* SGX-specific here. The > only question to me is whether any non-SGX users want something like > this and who they are. > > Here are some ideas I like more than an "sgx/" directory: > > /sys/devices/system/node/nodeX/arch/sgx_size > /sys/devices/system/node/nodeX/x86/sgx_size > /sys/devices/system/node/nodeX/coco/sgx_size Sure, I can rename the attribute group as "x86". > There's somebody else *today* who is trying to do something in the same > general area: per-node platform-specific memory encryption capabilities: > > > https://lore.kernel.org/all/20211027195511.207552-6-martin.fernandez@eclypsium.com/ Martin's is adding a new attribute as part of pre-existing attribute group for node device, where as my patch is adding a completely new named attribute group. > Also, could we please think through how this will look if we add more > attributes? I can imagine wanting both: > > * total SGX memory available > * total SGX memory present > > But those would be quite hard to differentiate if we have just an > "sgx_size". > > Wouldn't it be much nicer to name them things like: > > sgx_present_bytes > sgx_available_bytes > > ? > > In other words, can we please try to think just a bit into the future on > this one? What other SGX things will we want to export like this? > Outside of SGX, who else wants stuff _like_ this? I don't mind renaming the attribute but maybe it should sgx_total_bytes, just because in some other sysfs attributes that keyword is used to refer all of the bytes? /Jarkko
On Sun, Oct 31, 2021 at 07:39:32PM +0200, Jarkko Sakkinen wrote: > > > +static ssize_t size_show(struct device *dev, struct device_attribute *attr, char *buf) > > > +{ > > > + unsigned long size = 0; > > > + int nid; > > > + > > > + for (nid = 0; nid < num_possible_nodes(); nid++) { > > > + if (dev == sgx_numa_nodes[nid].dev) { > > > + size = sgx_numa_nodes[nid].size; > > > + break; > > > + } > > > + } > > > + > > > + return sysfs_emit(buf, "%lu\n", size); > > > +} > > > +DEVICE_ATTR_RO(size); > > > > static? > > The named ("sgx") attribute group is exported: > > extern const struct attribute_group sgx_node_group; That's fine, I am objecting to the fact that you now have added a global symbol called device_attr_size. Please make that static as that is a VERY generic name. thanks, greg k-h
On Sun, 2021-10-31 at 20:30 +0100, Greg Kroah-Hartman wrote: > On Sun, Oct 31, 2021 at 07:39:32PM +0200, Jarkko Sakkinen wrote: > > > > +static ssize_t size_show(struct device *dev, struct device_attribute *attr, char *buf) > > > > +{ > > > > + unsigned long size = 0; > > > > + int nid; > > > > + > > > > + for (nid = 0; nid < num_possible_nodes(); nid++) { > > > > + if (dev == sgx_numa_nodes[nid].dev) { > > > > + size = sgx_numa_nodes[nid].size; > > > > + break; > > > > + } > > > > + } > > > > + > > > > + return sysfs_emit(buf, "%lu\n", size); > > > > +} > > > > +DEVICE_ATTR_RO(size); > > > > > > static? > > > > The named ("sgx") attribute group is exported: > > > > extern const struct attribute_group sgx_node_group; > > That's fine, I am objecting to the fact that you now have added a global > symbol called device_attr_size. Please make that static as that is a > VERY generic name. Right, got it! Thank you. /Jarkko
On Sun, 2021-10-31 at 19:49 +0200, Jarkko Sakkinen wrote: > On Fri, 2021-10-29 at 10:00 -0700, Dave Hansen wrote: > > On 10/29/21 5:18 AM, Jarkko Sakkinen wrote: > > > +What: /sys/devices/system/node/nodeX/sgx/size > > > +Date: October 2021 > > > +Contact: Jarkko Sakkinen <jarkko@kernel.org> > > > > I don't think we should do something *entirely* SGX-specific here. The > > only question to me is whether any non-SGX users want something like > > this and who they are. > > > > Here are some ideas I like more than an "sgx/" directory: > > > > /sys/devices/system/node/nodeX/arch/sgx_size > > /sys/devices/system/node/nodeX/x86/sgx_size > > /sys/devices/system/node/nodeX/coco/sgx_size > > Sure, I can rename the attribute group as "x86". Dave, So, if I rename sgx_node_group as arch_node_dev_group, we also need something like CONFIG_ARCH_HAS_NODE_DEV_GROUP defined in arch/Kconfig, so that the group definition in drivers/base/node.c becomes: static const struct attribute_group *node_dev_groups[] = { &node_dev_group, #ifdef CONFIG_ARCH_HAS_NODE_DEV_GROUP &arch_node_dev_group, #endif NULL, } Does this make sense to you, and reflect your feedback in a reasonable manner? /Jarkko
diff --git a/Documentation/ABI/stable/sysfs-devices-node b/Documentation/ABI/stable/sysfs-devices-node index 484fc04bcc25..12dc2149e8e0 100644 --- a/Documentation/ABI/stable/sysfs-devices-node +++ b/Documentation/ABI/stable/sysfs-devices-node @@ -176,3 +176,10 @@ Contact: Keith Busch <keith.busch@intel.com> Description: The cache write policy: 0 for write-back, 1 for write-through, other or unknown. + +What: /sys/devices/system/node/nodeX/sgx/size +Date: October 2021 +Contact: Jarkko Sakkinen <jarkko@kernel.org> +Description: + Total available physical SGX memory, also known as Enclave Page + Cache (EPC), in bytes. diff --git a/arch/x86/include/asm/sgx.h b/arch/x86/include/asm/sgx.h index 05f3e21f01a7..43d1ccae6c62 100644 --- a/arch/x86/include/asm/sgx.h +++ b/arch/x86/include/asm/sgx.h @@ -365,6 +365,8 @@ struct sgx_sigstruct { * comment! */ +extern const struct attribute_group sgx_node_group; + #ifdef CONFIG_X86_SGX_KVM int sgx_virt_ecreate(struct sgx_pageinfo *pageinfo, void __user *secs, int *trapnr); diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c index a6e313f1a82d..86d6d309ff29 100644 --- a/arch/x86/kernel/cpu/sgx/main.c +++ b/arch/x86/kernel/cpu/sgx/main.c @@ -714,9 +714,12 @@ static bool __init sgx_page_cache_init(void) spin_lock_init(&sgx_numa_nodes[nid].lock); INIT_LIST_HEAD(&sgx_numa_nodes[nid].free_page_list); node_set(nid, sgx_numa_mask); + sgx_numa_nodes[nid].size = 0; } sgx_epc_sections[i].node = &sgx_numa_nodes[nid]; + sgx_numa_nodes[nid].dev = &node_devices[nid]->dev; + sgx_numa_nodes[nid].size += size; sgx_nr_epc_sections++; } @@ -790,6 +793,34 @@ int sgx_set_attribute(unsigned long *allowed_attributes, } EXPORT_SYMBOL_GPL(sgx_set_attribute); +#ifdef CONFIG_NUMA +static ssize_t size_show(struct device *dev, struct device_attribute *attr, char *buf) +{ + unsigned long size = 0; + int nid; + + for (nid = 0; nid < num_possible_nodes(); nid++) { + if (dev == sgx_numa_nodes[nid].dev) { + size = sgx_numa_nodes[nid].size; + break; + } + } + + return sysfs_emit(buf, "%lu\n", size); +} +DEVICE_ATTR_RO(size); + +static struct attribute *sgx_node_attrs[] = { + &dev_attr_size.attr, + NULL, +}; + +const struct attribute_group sgx_node_group = { + .name = "sgx", + .attrs = sgx_node_attrs, +}; +#endif /* CONFIG_NUMA */ + static int __init sgx_init(void) { int ret; diff --git a/arch/x86/kernel/cpu/sgx/sgx.h b/arch/x86/kernel/cpu/sgx/sgx.h index 4628acec0009..1de8c627a286 100644 --- a/arch/x86/kernel/cpu/sgx/sgx.h +++ b/arch/x86/kernel/cpu/sgx/sgx.h @@ -39,6 +39,8 @@ struct sgx_epc_page { */ struct sgx_numa_node { struct list_head free_page_list; + struct device *dev; + unsigned long size; spinlock_t lock; }; diff --git a/drivers/base/node.c b/drivers/base/node.c index 4a4ae868ad9f..3f85131171f0 100644 --- a/drivers/base/node.c +++ b/drivers/base/node.c @@ -20,6 +20,9 @@ #include <linux/pm_runtime.h> #include <linux/swap.h> #include <linux/slab.h> +#ifdef CONFIG_X86_SGX +#include <asm/sgx.h> +#endif static struct bus_type node_subsys = { .name = "node", @@ -565,7 +568,18 @@ static struct attribute *node_dev_attrs[] = { &dev_attr_vmstat.attr, NULL }; -ATTRIBUTE_GROUPS(node_dev); + +static const struct attribute_group node_dev_group = { + .attrs = node_dev_attrs, +}; + +static const struct attribute_group *node_dev_groups[] = { + &node_dev_group, +#ifdef CONFIG_X86_SGX + &sgx_node_group, +#endif /* CONFIG_X86_SGX */ + NULL, +}; #ifdef CONFIG_HUGETLBFS /*
The amount of SGX memory on the system is determined by the BIOS and it varies wildly between systems. It can be from dozens of MB's on desktops or VM's, up to many GB's on servers. Just like for regular memory, it is sometimes useful to know the amount of usable SGX memory in the system. Add an attribute for the amount of SGX memory in bytes to each NUMA node. The path is /sys/devices/system/node/node[0-9]*/sgx/size. Calculate these values by summing up EPC section sizes for each node during the driver initalization. Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org> --- v9: * Fix racy initialization of sysfs attributes: https://lore.kernel.org/linux-sgx/YXOsx8SvFJV5R7lU@kroah.com/ v8: * Fix a bug in sgx_numa_init(): node->dev should be only set after sysfe_create_group(). Otherwise, sysfs_remove_group() will issue a warning in sgx_numa_exit(), when sgx_create_group() is unsuccessful, because the group does not exist. v7: * Shorten memory_size to size. The prefix makes the name only longer but does not clarify things more than "size" would. * Use device_attribute instead of kobj_attribute. * Use named attribute group instead of creating raw kobject just for the "sgx" subdirectory. v6: * Initialize node->size to zero in sgx_setup_epc_section(), when the node is first accessed. v5 * A new patch based on the discussion on https://lore.kernel.org/linux-sgx/3a7cab4115b4f902f3509ad8652e616b91703e1d.camel@kernel.org/T/#t --- Documentation/ABI/stable/sysfs-devices-node | 7 +++++ arch/x86/include/asm/sgx.h | 2 ++ arch/x86/kernel/cpu/sgx/main.c | 31 +++++++++++++++++++++ arch/x86/kernel/cpu/sgx/sgx.h | 2 ++ drivers/base/node.c | 16 ++++++++++- 5 files changed, 57 insertions(+), 1 deletion(-)