- 18 Dec, 2020 40 commits
-
-
Max Gurtovoy authored
Introduce new vdpa_sim_net and vdpa_sim (core) drivers. This is a preparation for adding a vdpa simulator module for block devices. Signed-off-by:
Max Gurtovoy <mgurtovoy@nvidia.com> [sgarzare: various cleanups/fixes] Acked-by:
Jason Wang <jasowang@redhat.com> Signed-off-by:
Stefano Garzarella <sgarzare@redhat.com> Link: https://lore.kernel.org/r/20201215144256.155342-19-sgarzare@redhat.comSigned-off-by:
Michael S. Tsirkin <mst@redhat.com>
-
Stefano Garzarella authored
vringh_getdesc_iotlb() manages 2 iovs for writable and readable descriptors. This is very useful for the block device, where for each request we have both types of descriptor. Let's split the vdpasim_virtqueue's iov field in out_iov and in_iov to use them with vringh_getdesc_iotlb(). We are using VIRTIO terminology for "out" (readable by the device) and "in" (writable by the device) descriptors. Acked-by:
-
Stefano Garzarella authored
Allow each device to specify the size of the buffer allocated in vdpa_sim. Acked-by:
-
Stefano Garzarella authored
The next patch will make the buffer size configurable from each device. Since the buffer could be larger than a page, we use kvmalloc() instead of kmalloc(). Acked-by:
-
Stefano Garzarella authored
Instead of calling the vq callback directly, we can leverage the vringh_notify() function, adding vdpasim_vq_notify() and setting it in the vringh notify callback. Suggested-by:
-
Stefano Garzarella authored
The set_config callback can be used by the device to parse the config structure modified by the driver. The callback will be invoked, if set, in vdpasim_set_config() after copying bytes from caller buffer into vdpasim->config buffer. Acked-by:
-
Stefano Garzarella authored
The get_config callback can be used by the device to fill the config structure. The callback will be invoked in vdpasim_get_config() before copying bytes into caller buffer. Move vDPA-net config updates from vdpasim_set_features() in the new vdpasim_net_get_config() callback. This is safe since in vdpa_get_config() we already check that .set_features() callback is called before .get_config(). Signed-off-by:
-
Stefano Garzarella authored
Add new 'config_size' attribute in 'vdpasim_dev_attr' and allocates 'config' dynamically to support any device types. Acked-by:
-
Stefano Garzarella authored
As preparation for the next patches, we store the MAC address, parsed during the vdpasim_create(), in a buffer that will be used to fill 'config' together with other configurations. Acked-by:
-
Stefano Garzarella authored
Rename vdpasim_work() in vdpasim_net_work() and add it to the vdpasim_dev_attr structure. Co-developed-by:
-
Stefano Garzarella authored
Introduce a new VDPASIM_FEATURES macro with the generic features supported by the vDPA simulator, and VDPASIM_NET_FEATURES macro with vDPA-net features. Add 'supported_features' field in vdpasim_dev_attr, to allow devices to specify their features. Co-developed-by:
-
Stefano Garzarella authored
Remove VDPASIM_DEVICE_ID macro and add 'id' field in vdpasim_dev_attr, that will be returned by vdpasim_get_device_id(). Use VIRTIO_ID_NET for vDPA-net simulator device id. Co-developed-by:
-
Stefano Garzarella authored
vdpasim_dev_attr will contain device specific attributes. We starting moving the number of virtqueues (i.e. nvqs) to vdpasim_dev_attr. vdpasim_create() creates a new vDPA simulator following the device attributes defined in the vdpasim_dev_attr parameter. Co-developed-by:
-
Stefano Garzarella authored
These variables store generic callbacks used by the vDPA simulator core, so we can remove the 'net' word in their names. Co-developed-by:
-
Stefano Garzarella authored
Some devices may require a higher limit for the number of IOTLB entries, so let's make it configurable through a module parameter. By default, it's initialized with the current limit (2048). Suggested-by:
-
Max Gurtovoy authored
Add a new attribute that will define the number of virt queues to be created for the vdpasim device. Signed-off-by:
-
Stefano Garzarella authored
Some headers are not necessary, so let's remove them to do some cleaning. Signed-off-by:
-
Stefano Garzarella authored
'default n' is not necessary since it is already the default when nothing is specified. Suggested-by:
-
Christophe JAILLET authored
'pci_set_dma_mask()' + 'pci_set_consistent_dma_mask()' can be replaced by an equivalent 'dma_set_mask_and_coherent()' which is much less verbose. While at it, fix a typo (s/confiugration/configuration) Signed-off-by:
Christophe JAILLET <christophe.jaillet@wanadoo.fr> Link: https://lore.kernel.org/r/20201129125434.1462638-1-christophe.jaillet@wanadoo.frSigned-off-by:
-
Tian Tao authored
Replace opencoded alloc and copy with vmemdup_user() Signed-off-by:
Tian Tao <tiantao6@hisilicon.com> Link: https://lore.kernel.org/r/1605057288-60400-1-git-send-email-tiantao6@hisilicon.comSigned-off-by:
-
David Hildenbrand authored
Let's add a safe mechanism to unplug memory, avoiding long/endless loops when trying to offline memory - similar to in SBM. Fake-offline all memory (via alloc_contig_range()) before trying to offline+remove it. Use this mode as default, but allow to enable the other mode explicitly (which could give better memory hotunplug guarantees in some environments). The "unsafe" mode can be enabled e.g., via virtio_mem.bbm_safe_unplug=0 on the cmdline. Reviewed-by:
Wei Yang <richard.weiyang@linux.alibaba.com> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Jason Wang <jasowang@redhat.com> Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Oscar Salvador <osalvador@suse.de> Cc: Wei Yang <richard.weiyang@linux.alibaba.com> Cc: Andrew Morton <akpm@linux-foundation.org> Signed-off-by:
David Hildenbrand <david@redhat.com> Link: https://lore.kernel.org/r/20201112133815.13332-30-david@redhat.comSigned-off-by:
-
David Hildenbrand authored
Let's try to unplug completely offline big blocks first. Then, (if enabled via unplug_offline) try to offline and remove whole big blocks. No locking necessary - we can deal with concurrent onlining/offlining just fine. Note1: This is sub-optimal and might be dangerous in some environments: we could end up in an infinite loop when offlining (e.g., long-term pinnings), similar as with DIMMs. We'll introduce safe memory hotunplug via fake-offlining next, and use this basic mode only when explicitly enabled. Note2: Without ZONE_MOVABLE, memory unplug will be extremely unreliable with bigger block sizes. Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Jason Wang <jasowang@redhat.com> Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Oscar Salvador <osalvador@suse.de> Cc: Wei Yang <richard.weiyang@linux.alibaba.com> Cc: Andrew Morton <akpm@linux-foundation.org> Signed-off-by:
-
David Hildenbrand authored
virtio-mem soon wants to use offline_and_remove_memory() memory that exceeds a single Linux memory block (memory_block_size_bytes()). Let's remove that restriction. Let's remember the old state and try to restore that if anything goes wrong. While re-onlining can, in general, fail, it's highly unlikely to happen (usually only when a notifier fails to allocate memory, and these are rather rare). This will be used by virtio-mem to offline+remove memory ranges that are bigger than a single memory block - for example, with a device block size of 1 GiB (e.g., gigantic pages in the hypervisor) and a Linux memory block size of 128MB. While we could compress the state into 2 bit, using 8 bit is much easier. This handling is similar, but different to acpi_scan_try_to_offline(): a) We don't try to offline twice. I am not sure if this CONFIG_MEMCG optimization is still relevant - it should only apply to ZONE_NORMAL (where we have no guarantees). If relevant, we can always add it. b) acpi_scan_try_to_offline() simply onlines all memory in case something goes wrong. It doesn't restore previous online type. Let's do that, so we won't overwrite what e.g., user space configured. Reviewed-by:
Andrew Morton <akpm@linux-foundation.org>
-
David Hildenbrand authored
Let's allow to force BBM, even if subblocks would be possible. Take care of properly calculating the first big block id, because the start address might no longer be aligned to the big block size. Also, allow to manually configure the size of Big Blocks. Reviewed-by:
-
David Hildenbrand authored
Currently, we do not support device block sizes that exceed the Linux memory block size. For example, having a device block size of 1 GiB (e.g., gigantic pages in the hypervisor) won't work with 128 MiB Linux memory blocks. Let's implement Big Block Mode (BBM), whereby we add/remove at least one Linux memory block at a time. With a 1 GiB device block size, a Big Block (BB) will cover 8 Linux memory blocks. We'll keep registering the online_page_callback machinery, it will be used for safe memory hotunplug in BBM next. Note: BBM is properly prepared for variable-sized Linux memory blocks that we might see in the future. So we won't care how many Linux memory blocks a big block actually spans, and how the memory notifier is called. Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Jason Wang <jasowang@redhat.com> Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Oscar Salvador <osalvador@suse.de> Cc: Wei Yang <richard.weiyang@linux.alibaba.com> Cc: Andrew Morton <akpm@linux-foundation.org> Signed-off-by:
-
David Hildenbrand authored
Let's use wrappers for the low-level functions that dev_dbg/dev_warn and work on addr + size, such that we can reuse them for adding/removing in other granularity. We only warn when adding memory failed, because that's something to pay attention to. We won't warn when removing failed, we'll reuse that in racy context soon (and we do have proper BUG_ON() statements in the current cases where it must never happen). Reviewed-by:
-
David Hildenbrand authored
Let's rename accordingly. Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Jason Wang <jasowang@redhat.com> Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com> Signed-off-by:
-
David Hildenbrand authored
Let's rename them accordingly. virtio_mem_plug_request() and virtio_mem_unplug_request() will be handled separately. Reviewed-by:
-
David Hildenbrand authored
Let's move first_mb_id/next_mb_id/last_usable_mb_id accordingly. Reviewed-by:
-
David Hildenbrand authored
Let's rename to "sbs_per_mb" and "sb_size" and move accordingly. Reviewed-by:
-
David Hildenbrand authored
Let's rename and move accordingly. While at it, rename sb_bitmap to "sb_states". Reviewed-by:
Pankaj Gupta <pankaj.gupta.linux@gmail.com> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Jason Wang <jasowang@redhat.com> Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com> Signed-off-by:
-
David Hildenbrand authored
let's use a new "sbm" sub-struct to hold SBM-specific state and rename + move applicable definitions, functions, and variables (related to memory block states). While at it: - Drop the "_STATE" part from memory block states - Rename "nb_mb_state" to "mb_count" - "set_mb_state" / "get_mb_state" vs. "mb_set_state" / "mb_get_state" - Don't use lengthy "enum virtio_mem_smb_mb_state", simply use "uint8_t" Reviewed-by:
-
David Hildenbrand authored
Let's add some documentation for the current mode - Sub Block Mode (SBM) - to prepare for a new mode - Big Block Mode (BBM). Follow-up patches will properly factor out the existing Sub Block Mode (SBM) and implement Big Block Mode (BBM). Reviewed-by:
-
David Hildenbrand authored
We don't want to add too much memory when it's not getting onlined immediately, to avoid running OOM. Generalize the handling, to avoid making use of memory block states. Use a threshold of 1 GiB for now. Properly adjust the offline size when adding/removing memory. As we are not always protected by a lock when touching the offline size, use an atomic64_t. We don't care about races (e.g., someone offlining memory while we are adding more), only about consistent values. (1 GiB needs a memmap of ~16MiB - which sounds reasonable even for setups with little boot memory and (possibly) one virtio-mem device per node) We don't want to retrigger when onlining is caused immediately by our action (e.g., adding memory which immediately gets onlined), so use a flag to indicate if the workqueue is active and use that as an indicator whether to trigger a retry. This will also be especially relevant for Big Block Mode (BBM), whereby we might re-online memory in case offlining of another memory block failed. Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Jason Wang <jasowang@redhat.com> Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com> Signed-off-by:
-
David Hildenbrand authored
Let's trigger from offlining code only when we're not allowed to unplug online memory. Handle the other case (memmap possibly freeing up another memory block) when actually removing memory. We now also properly handle the case when removing already offline memory blocks via virtio_mem_mb_remove(). When removing via virtio_mem_remove(), when unloading the driver, virtio_mem_retry() is a NOP and safe to use. While at it, move retry handling when offlining out of virtio_mem_notify_offline(), to share it with Big Block Mode (BBM) soon. This is a preparation for Big Block Mode (BBM), whereby we can see some temporary offlining of memory blocks without actually making progress. Imagine you have a Big Block that spans to Linux memory blocks. Assume the first Linux memory blocks has no unmovable data on it. When we would call offline_and_remove_memory() on the big block, we would 1. Try to offline the first block. Works, notifiers triggered. virtio_mem_retry() called. 2. Try to offline the second block. Does not work. 3. Re-online first block. 4. Exit to main loop, exit workqueue. 5. Retry immediately (due to virtio_mem_retry()), go to 1. The result are endless retries. Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Jason Wang <jasowang@redhat.com> Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com> Signed-off-by:
-
David Hildenbrand authored
No longer used, let's drop it. Reviewed-by:
-
David Hildenbrand authored
Avoid using memory block ids. While at it, use uint64_t for address/size. This is a preparation for Big Block Mode (BBM). Reviewed-by:
-
David Hildenbrand authored
Avoid using memory block ids. Rename it to virtio_mem_contains_range(). This is a preparation for Big Block Mode (BBM). Reviewed-by:
-
David Hildenbrand authored
Let's check by traversing busy system RAM resources instead, to avoid relying on memory block states. Don't use walk_system_ram_range(), as that works on pages and we want to use the bare addresses we have easily at hand. This is a preparation for Big Block Mode (BBM), which won't have memory block states. Reviewed-by:
-
David Hildenbrand authored
ZONE_MOVABLE is supposed to give some guarantees, yet, alloc_contig_range() isn't prepared to properly deal with some racy cases properly (e.g., temporary page pinning when exiting processed, PCP). Retry 5 times for now. There is certainly room for improvement in the future. Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Jason Wang <jasowang@redhat.com> Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com> Signed-off-by:
-
