Linaro-mm-sig

linaro-mm-sig@lists.linaro.org

15 participants
3163 discussions

[PATCH 10/15] serial: ambarella: add support for Ambarella uart_port

by lchen＠ambarella.com

The content of this message was lost. It was probably cross-posted to multiple lists and previously handled on another list.

2 years, 9 months

[PATCH 0/4] Track exported dma-buffers with memcg

by T.J. Mercier

Based on discussions at LPC, this series adds a memory.stat counter for exported dmabufs. This counter allows us to continue tracking system-wide total exported buffer sizes which there is no longer any way to get without DMABUF_SYSFS_STATS, and adds a new capability to track per-cgroup exported buffer sizes. The total (root counter) is helpful for accounting in-kernel dmabuf use (by comparing with the sum of child nodes or with the sum of sizes of mapped buffers or FD references in procfs) in addition to helping identify driver memory leaks when in-kernel use continually increases over time. With per-application cgroups, the per-cgroup counter allows us to quickly see how much dma-buf memory an application has caused to be allocated. This avoids the need to read through all of procfs which can be a lengthy process, and causes the charge to "stick" to the allocating process/cgroup as long as the buffer is alive, regardless of how the buffer is shared (unless the charge is transferred). The first patch adds the counter to memcg. The next two patches allow the charge for a buffer to be transferred across cgroups which is necessary because of the way most dmabufs are allocated from a central process on Android. The fourth patch adds a SELinux hook to binder in order to control who is allowed to transfer buffer charges. [1] https://lore.kernel.org/all/20220617085702.4298-1-christian.koenig@amd.com/ Hridya Valsaraju (1): binder: Add flags to relinquish ownership of fds T.J. Mercier (3): memcg: Track exported dma-buffers dmabuf: Add cgroup charge transfer function security: binder: Add transfer_charge SElinux hook Documentation/admin-guide/cgroup-v2.rst | 5 +++ drivers/android/binder.c | 36 +++++++++++++++-- drivers/dma-buf/dma-buf.c | 54 +++++++++++++++++++++++-- include/linux/dma-buf.h | 5 +++ include/linux/lsm_hook_defs.h | 2 + include/linux/lsm_hooks.h | 6 +++ include/linux/memcontrol.h | 7 ++++ include/linux/security.h | 2 + include/uapi/linux/android/binder.h | 23 +++++++++-- mm/memcontrol.c | 4 ++ security/security.c | 6 +++ security/selinux/hooks.c | 9 +++++ security/selinux/include/classmap.h | 2 +- 13 files changed, 149 insertions(+), 12 deletions(-) base-commit: b7bfaa761d760e72a969d116517eaa12e404c262 -- 2.39.0.314.g84b9a713c41-goog

2 years, 10 months

[PATCH v2] drm/msm: Add missing check and destroy for alloc_ordered_workqueue

by Jiasheng Jiang

Add check for the return value of alloc_ordered_workqueue as it may return NULL pointer. Moreover, use the destroy_workqueue in the later fails in order to avoid memory leak. Signed-off-by: Jiasheng Jiang <jiasheng(a)iscas.ac.cn> --- Changelog: v1 -> v2: 1. Convert "goto err_destroy_workqueue" into "goto err_msm_unit" and remove "err_destroy_workqueue" label. --- drivers/gpu/drm/msm/msm_drv.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/msm/msm_drv.c b/drivers/gpu/drm/msm/msm_drv.c index 8b0b0ac74a6f..54be323ed33d 100644 --- a/drivers/gpu/drm/msm/msm_drv.c +++ b/drivers/gpu/drm/msm/msm_drv.c @@ -418,6 +418,8 @@ static int msm_drm_init(struct device *dev, const struct drm_driver *drv) priv->dev = ddev; priv->wq = alloc_ordered_workqueue("msm", 0); + if (!priv->wq) + return -ENOMEM; INIT_LIST_HEAD(&priv->objects); mutex_init(&priv->obj_lock); @@ -440,12 +442,12 @@ static int msm_drm_init(struct device *dev, const struct drm_driver *drv) ret = msm_init_vram(ddev); if (ret) - return ret; + goto err_msm_uninit; /* Bind all our sub-components: */ ret = component_bind_all(dev, ddev); if (ret) - return ret; + goto err_msm_uninit; dma_set_max_seg_size(dev, UINT_MAX); -- 2.25.1

2 years, 10 months

[PATCH] drm/msm/gem: Add check for kmalloc

by Jiasheng Jiang

Add the check for the return value of kmalloc in order to avoid NULL pointer dereference in copy_from_user. Fixes: 20224d715a88 ("drm/msm/submit: Move copy_from_user ahead of locking bos") Signed-off-by: Jiasheng Jiang <jiasheng(a)iscas.ac.cn> --- drivers/gpu/drm/msm/msm_gem_submit.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/drivers/gpu/drm/msm/msm_gem_submit.c b/drivers/gpu/drm/msm/msm_gem_submit.c index 45a3e5cadc7d..7c2cc1262c05 100644 --- a/drivers/gpu/drm/msm/msm_gem_submit.c +++ b/drivers/gpu/drm/msm/msm_gem_submit.c @@ -209,6 +209,10 @@ static int submit_lookup_cmds(struct msm_gem_submit *submit, goto out; } submit->cmd[i].relocs = kmalloc(sz, GFP_KERNEL); + if (!submit->cmd[i].relocs) { + ret = -ENOMEM; + goto out; + } ret = copy_from_user(submit->cmd[i].relocs, userptr, sz); if (ret) { ret = -EFAULT; -- 2.25.1

2 years, 10 months

[PATCH RESEND] drm/msm: Add missing check and destroy for alloc_ordered_workqueue

by Jiasheng Jiang

Add check for the return value of alloc_ordered_workqueue as it may return NULL pointer. Moreover, use the destroy_workqueue in the later fails in order to avoid memory leak. Fixes: c8afe684c95c ("drm/msm: basic KMS driver for snapdragon") Signed-off-by: Jiasheng Jiang <jiasheng(a)iscas.ac.cn> --- drivers/gpu/drm/msm/msm_drv.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/msm/msm_drv.c b/drivers/gpu/drm/msm/msm_drv.c index 8b0b0ac74a6f..b82d938226ad 100644 --- a/drivers/gpu/drm/msm/msm_drv.c +++ b/drivers/gpu/drm/msm/msm_drv.c @@ -418,6 +418,8 @@ static int msm_drm_init(struct device *dev, const struct drm_driver *drv) priv->dev = ddev; priv->wq = alloc_ordered_workqueue("msm", 0); + if (!priv->wq) + return -ENOMEM; INIT_LIST_HEAD(&priv->objects); mutex_init(&priv->obj_lock); @@ -440,12 +442,12 @@ static int msm_drm_init(struct device *dev, const struct drm_driver *drv) ret = msm_init_vram(ddev); if (ret) - return ret; + goto err_destroy_workqueue; /* Bind all our sub-components: */ ret = component_bind_all(dev, ddev); if (ret) - return ret; + goto err_destroy_workqueue; dma_set_max_seg_size(dev, UINT_MAX); @@ -540,6 +542,8 @@ static int msm_drm_init(struct device *dev, const struct drm_driver *drv) err_msm_uninit: msm_drm_uninit(dev); +err_destroy_workqueue: + destroy_workqueue(priv->wq); return ret; } -- 2.25.1

2 years, 10 months

[PATCH] drm/msm: Add missing check and destroy for alloc_ordered_workqueue

by Jiasheng Jiang

2 years, 10 months

Try to address the DMA-buf coherency problem

by Christian König

Hi guys, after finding that we essentially have two separate worlds for coherent sharing of buffer through DMA-buf I thought I will tackle that problem a bit and at least allow the framework to reject attachments which won't work. So those patches here add a new dma_coherent flag to each DMA-buf object telling the framework that dev_is_dma_coherent() needs to return true for an importing device to be able to attach. Since we should always have a fallback path this should give userspace the chance to still keep the use case working, either by doing a CPU copy instead or reversing the roles of exporter and importer. For DRM and most V4L2 devices I then fill in the dma_coherent flag based on the return value of dev_is_dma_coherent(). Exporting drivers are allowed to clear the flag for their buffers if special handling like the USWC flag in amdgpu or the uncached allocations for radeon/nouveau are in use. Additional to that importers can also check the flag if they have some non-snooping operations like the special scanout case for amdgpu for example. The patches are only smoke tested and the solution isn't ideal, but as far as I can see should at least keep things working. Please review and/or comment, Christian.

2 years, 10 months

[PATCH] dma-buf: fix dma_buf_export init order v2

by Christian König

The init order and resulting error handling in dma_buf_export was pretty messy. Subordinate objects like the file and the sysfs kernel objects were initializing and wiring itself up with the object in the wrong order resulting not only in complicating and partially incorrect error handling, but also in publishing only halve initialized DMA-buf objects. Clean this up thoughtfully by allocating the file independent of the DMA-buf object. Then allocate and initialize the DMA-buf object itself, before publishing it through sysfs. If everything works as expected the file is then connected with the DMA-buf object and publish it through debugfs. Also adds the missing dma_resv_fini() into the error handling. v2: add some missing changes to dma_bug_getfile() and a missing NULL check in dma_buf_file_release() Signed-off-by: Christian König <christian.koenig(a)amd.com> --- drivers/dma-buf/dma-buf-sysfs-stats.c | 7 +-- drivers/dma-buf/dma-buf-sysfs-stats.h | 4 +- drivers/dma-buf/dma-buf.c | 84 +++++++++++++-------------- 3 files changed, 43 insertions(+), 52 deletions(-) diff --git a/drivers/dma-buf/dma-buf-sysfs-stats.c b/drivers/dma-buf/dma-buf-sysfs-stats.c index 2bba0babcb62..4b680e10c15a 100644 --- a/drivers/dma-buf/dma-buf-sysfs-stats.c +++ b/drivers/dma-buf/dma-buf-sysfs-stats.c @@ -168,14 +168,11 @@ void dma_buf_uninit_sysfs_statistics(void) kset_unregister(dma_buf_stats_kset); } -int dma_buf_stats_setup(struct dma_buf *dmabuf) +int dma_buf_stats_setup(struct dma_buf *dmabuf, struct file *file) { struct dma_buf_sysfs_entry *sysfs_entry; int ret; - if (!dmabuf || !dmabuf->file) - return -EINVAL; - if (!dmabuf->exp_name) { pr_err("exporter name must not be empty if stats needed\n"); return -EINVAL; @@ -192,7 +189,7 @@ int dma_buf_stats_setup(struct dma_buf *dmabuf) /* create the directory for buffer stats */ ret = kobject_init_and_add(&sysfs_entry->kobj, &dma_buf_ktype, NULL, - "%lu", file_inode(dmabuf->file)->i_ino); + "%lu", file_inode(file)->i_ino); if (ret) goto err_sysfs_dmabuf; diff --git a/drivers/dma-buf/dma-buf-sysfs-stats.h b/drivers/dma-buf/dma-buf-sysfs-stats.h index a49c6e2650cc..7a8a995b75ba 100644 --- a/drivers/dma-buf/dma-buf-sysfs-stats.h +++ b/drivers/dma-buf/dma-buf-sysfs-stats.h @@ -13,7 +13,7 @@ int dma_buf_init_sysfs_statistics(void); void dma_buf_uninit_sysfs_statistics(void); -int dma_buf_stats_setup(struct dma_buf *dmabuf); +int dma_buf_stats_setup(struct dma_buf *dmabuf, struct file *file); void dma_buf_stats_teardown(struct dma_buf *dmabuf); #else @@ -25,7 +25,7 @@ static inline int dma_buf_init_sysfs_statistics(void) static inline void dma_buf_uninit_sysfs_statistics(void) {} -static inline int dma_buf_stats_setup(struct dma_buf *dmabuf) +static inline int dma_buf_stats_setup(struct dma_buf *dmabuf, struct file *file) { return 0; } diff --git a/drivers/dma-buf/dma-buf.c b/drivers/dma-buf/dma-buf.c index e6f36c014c4c..eb6b59363c4f 100644 --- a/drivers/dma-buf/dma-buf.c +++ b/drivers/dma-buf/dma-buf.c @@ -95,10 +95,11 @@ static int dma_buf_file_release(struct inode *inode, struct file *file) return -EINVAL; dmabuf = file->private_data; - - mutex_lock(&db_list.lock); - list_del(&dmabuf->list_node); - mutex_unlock(&db_list.lock); + if (dmabuf) { + mutex_lock(&db_list.lock); + list_del(&dmabuf->list_node); + mutex_unlock(&db_list.lock); + } return 0; } @@ -523,17 +524,17 @@ static inline int is_dma_buf_file(struct file *file) return file->f_op == &dma_buf_fops; } -static struct file *dma_buf_getfile(struct dma_buf *dmabuf, int flags) +static struct file *dma_buf_getfile(size_t size, int flags) { static atomic64_t dmabuf_inode = ATOMIC64_INIT(0); - struct file *file; struct inode *inode = alloc_anon_inode(dma_buf_mnt->mnt_sb); + struct file *file; if (IS_ERR(inode)) return ERR_CAST(inode); - inode->i_size = dmabuf->size; - inode_set_bytes(inode, dmabuf->size); + inode->i_size = size; + inode_set_bytes(inode, size); /* * The ->i_ino acquired from get_next_ino() is not unique thus @@ -547,8 +548,6 @@ static struct file *dma_buf_getfile(struct dma_buf *dmabuf, int flags) flags, &dma_buf_fops); if (IS_ERR(file)) goto err_alloc_file; - file->private_data = dmabuf; - file->f_path.dentry->d_fsdata = dmabuf; return file; @@ -614,19 +613,11 @@ struct dma_buf *dma_buf_export(const struct dma_buf_export_info *exp_info) size_t alloc_size = sizeof(struct dma_buf); int ret; - if (!exp_info->resv) - alloc_size += sizeof(struct dma_resv); - else - /* prevent &dma_buf[1] == dma_buf->resv */ - alloc_size += 1; - - if (WARN_ON(!exp_info->priv - || !exp_info->ops - || !exp_info->ops->map_dma_buf - || !exp_info->ops->unmap_dma_buf - || !exp_info->ops->release)) { + if (WARN_ON(!exp_info->priv || !exp_info->ops + || !exp_info->ops->map_dma_buf + || !exp_info->ops->unmap_dma_buf + || !exp_info->ops->release)) return ERR_PTR(-EINVAL); - } if (WARN_ON(exp_info->ops->cache_sgt_mapping && (exp_info->ops->pin || exp_info->ops->unpin))) @@ -638,10 +629,21 @@ struct dma_buf *dma_buf_export(const struct dma_buf_export_info *exp_info) if (!try_module_get(exp_info->owner)) return ERR_PTR(-ENOENT); + file = dma_buf_getfile(exp_info->size, exp_info->flags); + if (IS_ERR(file)) { + ret = PTR_ERR(file); + goto err_module; + } + + if (!exp_info->resv) + alloc_size += sizeof(struct dma_resv); + else + /* prevent &dma_buf[1] == dma_buf->resv */ + alloc_size += 1; dmabuf = kzalloc(alloc_size, GFP_KERNEL); if (!dmabuf) { ret = -ENOMEM; - goto err_module; + goto err_file; } dmabuf->priv = exp_info->priv; @@ -653,44 +655,36 @@ struct dma_buf *dma_buf_export(const struct dma_buf_export_info *exp_info) init_waitqueue_head(&dmabuf->poll); dmabuf->cb_in.poll = dmabuf->cb_out.poll = &dmabuf->poll; dmabuf->cb_in.active = dmabuf->cb_out.active = 0; + mutex_init(&dmabuf->lock); + INIT_LIST_HEAD(&dmabuf->attachments); if (!resv) { - resv = (struct dma_resv *)&dmabuf[1]; - dma_resv_init(resv); + dmabuf->resv = (struct dma_resv *)&dmabuf[1]; + dma_resv_init(dmabuf->resv); + } else { + dmabuf->resv = resv; } - dmabuf->resv = resv; - file = dma_buf_getfile(dmabuf, exp_info->flags); - if (IS_ERR(file)) { - ret = PTR_ERR(file); + ret = dma_buf_stats_setup(dmabuf, file); + if (ret) goto err_dmabuf; - } + file->private_data = dmabuf; + file->f_path.dentry->d_fsdata = dmabuf; dmabuf->file = file; - mutex_init(&dmabuf->lock); - INIT_LIST_HEAD(&dmabuf->attachments); - mutex_lock(&db_list.lock); list_add(&dmabuf->list_node, &db_list.head); mutex_unlock(&db_list.lock); - ret = dma_buf_stats_setup(dmabuf); - if (ret) - goto err_sysfs; - return dmabuf; -err_sysfs: - /* - * Set file->f_path.dentry->d_fsdata to NULL so that when - * dma_buf_release() gets invoked by dentry_ops, it exits - * early before calling the release() dma_buf op. - */ - file->f_path.dentry->d_fsdata = NULL; - fput(file); err_dmabuf: + if (!resv) + dma_resv_fini(dmabuf->resv); kfree(dmabuf); +err_file: + fput(file); err_module: module_put(exp_info->owner); return ERR_PTR(ret); -- 2.34.1

2 years, 10 months

[PATCH v2] crypto: ccp - Allocate TEE ring and cmd buffer using DMA APIs

by Rijo Thomas

For AMD Secure Processor (ASP) to map and access TEE ring buffer, the ring buffer address sent by host to ASP must be a real physical address and the pages must be physically contiguous. In a virtualized environment though, when the driver is running in a guest VM, the pages allocated by __get_free_pages() may not be contiguous in the host (or machine) physical address space. Guests will see a guest (or pseudo) physical address and not the actual host (or machine) physical address. The TEE running on ASP cannot decipher pseudo physical addresses. It needs host or machine physical address. To resolve this problem, use DMA APIs for allocating buffers that must be shared with TEE. This will ensure that the pages are contiguous in host (or machine) address space. If the DMA handle is an IOVA, translate it into a physical address before sending it to ASP. This patch also exports two APIs (one for buffer allocation and another to free the buffer). This API can be used by AMD-TEE driver to share buffers with TEE. Fixes: 33960acccfbd ("crypto: ccp - add TEE support for Raven Ridge") Cc: Tom Lendacky <thomas.lendacky(a)amd.com> Cc: stable(a)vger.kernel.org Signed-off-by: Rijo Thomas <Rijo-john.Thomas(a)amd.com> Co-developed-by: Jeshwanth <JESHWANTHKUMAR.NK(a)amd.com> Signed-off-by: Jeshwanth <JESHWANTHKUMAR.NK(a)amd.com> Reviewed-by: Devaraj Rangasamy <Devaraj.Rangasamy(a)amd.com> --- v2: * Removed references to dma_buffer. * If psp_init() fails, clear reference to master device. * Handle gfp flags within psp_tee_alloc_buffer() instead of passing it as a function argument. * Added comments within psp_tee_alloc_buffer() to serve as future documentation. drivers/crypto/ccp/psp-dev.c | 13 ++-- drivers/crypto/ccp/tee-dev.c | 124 +++++++++++++++++++++++------------ drivers/crypto/ccp/tee-dev.h | 9 +-- include/linux/psp-tee.h | 49 ++++++++++++++ 4 files changed, 142 insertions(+), 53 deletions(-) diff --git a/drivers/crypto/ccp/psp-dev.c b/drivers/crypto/ccp/psp-dev.c index c9c741ac8442..380f5caaa550 100644 --- a/drivers/crypto/ccp/psp-dev.c +++ b/drivers/crypto/ccp/psp-dev.c @@ -161,13 +161,13 @@ int psp_dev_init(struct sp_device *sp) goto e_err; } - ret = psp_init(psp); - if (ret) - goto e_irq; - if (sp->set_psp_master_device) sp->set_psp_master_device(sp); + ret = psp_init(psp); + if (ret) + goto e_clear; + /* Enable interrupt */ iowrite32(-1, psp->io_regs + psp->vdata->inten_reg); @@ -175,7 +175,10 @@ int psp_dev_init(struct sp_device *sp) return 0; -e_irq: +e_clear: + if (sp->clear_psp_master_device) + sp->clear_psp_master_device(sp); + sp_free_psp_irq(psp->sp, psp); e_err: sp->psp_data = NULL; diff --git a/drivers/crypto/ccp/tee-dev.c b/drivers/crypto/ccp/tee-dev.c index 5c9d47f3be37..5c43e6e166f1 100644 --- a/drivers/crypto/ccp/tee-dev.c +++ b/drivers/crypto/ccp/tee-dev.c @@ -12,8 +12,9 @@ #include <linux/mutex.h> #include <linux/delay.h> #include <linux/slab.h> +#include <linux/dma-direct.h> +#include <linux/iommu.h> #include <linux/gfp.h> -#include <linux/psp-sev.h> #include <linux/psp-tee.h> #include "psp-dev.h" @@ -21,25 +22,73 @@ static bool psp_dead; +struct psp_tee_buffer *psp_tee_alloc_buffer(unsigned long size) +{ + struct psp_device *psp = psp_get_master_device(); + struct psp_tee_buffer *buf; + struct iommu_domain *dom; + + if (!psp || !size) + return NULL; + + buf = kzalloc(sizeof(*buf), GFP_KERNEL); + if (!buf) + return NULL; + + /* The pages allocated for PSP Trusted OS must be physically + * contiguous in host (or machine) address space. Therefore, + * use DMA API to allocate memory. + */ + + buf->vaddr = dma_alloc_coherent(psp->dev, size, &buf->dma, + GFP_KERNEL | __GFP_ZERO); + if (!buf->vaddr || !buf->dma) { + kfree(buf); + return NULL; + } + + buf->size = size; + + /* Check whether IOMMU is present. If present, convert IOVA to + * physical address. In the absence of IOMMU, the DMA address + * is actually the physical address. + */ + + dom = iommu_get_domain_for_dev(psp->dev); + if (dom) + buf->paddr = iommu_iova_to_phys(dom, buf->dma); + else + buf->paddr = buf->dma; + + return buf; +} +EXPORT_SYMBOL(psp_tee_alloc_buffer); + +void psp_tee_free_buffer(struct psp_tee_buffer *buf) +{ + struct psp_device *psp = psp_get_master_device(); + + if (!psp || !buf) + return; + + dma_free_coherent(psp->dev, buf->size, buf->vaddr, buf->dma); + + kfree(buf); +} +EXPORT_SYMBOL(psp_tee_free_buffer); + static int tee_alloc_ring(struct psp_tee_device *tee, int ring_size) { struct ring_buf_manager *rb_mgr = &tee->rb_mgr; - void *start_addr; if (!ring_size) return -EINVAL; - /* We need actual physical address instead of DMA address, since - * Trusted OS running on AMD Secure Processor will map this region - */ - start_addr = (void *)__get_free_pages(GFP_KERNEL, get_order(ring_size)); - if (!start_addr) + rb_mgr->ring_buf = psp_tee_alloc_buffer(ring_size); + if (!rb_mgr->ring_buf) { + dev_err(tee->dev, "ring allocation failed\n"); return -ENOMEM; - - memset(start_addr, 0x0, ring_size); - rb_mgr->ring_start = start_addr; - rb_mgr->ring_size = ring_size; - rb_mgr->ring_pa = __psp_pa(start_addr); + } mutex_init(&rb_mgr->mutex); return 0; @@ -49,15 +98,8 @@ static void tee_free_ring(struct psp_tee_device *tee) { struct ring_buf_manager *rb_mgr = &tee->rb_mgr; - if (!rb_mgr->ring_start) - return; - - free_pages((unsigned long)rb_mgr->ring_start, - get_order(rb_mgr->ring_size)); + psp_tee_free_buffer(rb_mgr->ring_buf); - rb_mgr->ring_start = NULL; - rb_mgr->ring_size = 0; - rb_mgr->ring_pa = 0; mutex_destroy(&rb_mgr->mutex); } @@ -81,35 +123,35 @@ static int tee_wait_cmd_poll(struct psp_tee_device *tee, unsigned int timeout, return -ETIMEDOUT; } -static -struct tee_init_ring_cmd *tee_alloc_cmd_buffer(struct psp_tee_device *tee) +static struct psp_tee_buffer *tee_alloc_cmd_buffer(struct psp_tee_device *tee) { + struct psp_tee_buffer *cmd_buffer; struct tee_init_ring_cmd *cmd; - cmd = kzalloc(sizeof(*cmd), GFP_KERNEL); - if (!cmd) + cmd_buffer = psp_tee_alloc_buffer(sizeof(*cmd)); + if (!cmd_buffer) return NULL; - cmd->hi_addr = upper_32_bits(tee->rb_mgr.ring_pa); - cmd->low_addr = lower_32_bits(tee->rb_mgr.ring_pa); - cmd->size = tee->rb_mgr.ring_size; + cmd = (struct tee_init_ring_cmd *)cmd_buffer->vaddr; + cmd->hi_addr = upper_32_bits(tee->rb_mgr.ring_buf->paddr); + cmd->low_addr = lower_32_bits(tee->rb_mgr.ring_buf->paddr); + cmd->size = tee->rb_mgr.ring_buf->size; dev_dbg(tee->dev, "tee: ring address: high = 0x%x low = 0x%x size = %u\n", cmd->hi_addr, cmd->low_addr, cmd->size); - return cmd; + return cmd_buffer; } -static inline void tee_free_cmd_buffer(struct tee_init_ring_cmd *cmd) +static inline void tee_free_cmd_buffer(struct psp_tee_buffer *cmd_buffer) { - kfree(cmd); + psp_tee_free_buffer(cmd_buffer); } static int tee_init_ring(struct psp_tee_device *tee) { int ring_size = MAX_RING_BUFFER_ENTRIES * sizeof(struct tee_ring_cmd); - struct tee_init_ring_cmd *cmd; - phys_addr_t cmd_buffer; + struct psp_tee_buffer *cmd_buffer; unsigned int reg; int ret; @@ -123,21 +165,19 @@ static int tee_init_ring(struct psp_tee_device *tee) tee->rb_mgr.wptr = 0; - cmd = tee_alloc_cmd_buffer(tee); - if (!cmd) { + cmd_buffer = tee_alloc_cmd_buffer(tee); + if (!cmd_buffer) { tee_free_ring(tee); return -ENOMEM; } - cmd_buffer = __psp_pa((void *)cmd); - /* Send command buffer details to Trusted OS by writing to * CPU-PSP message registers */ - iowrite32(lower_32_bits(cmd_buffer), + iowrite32(lower_32_bits(cmd_buffer->paddr), tee->io_regs + tee->vdata->cmdbuff_addr_lo_reg); - iowrite32(upper_32_bits(cmd_buffer), + iowrite32(upper_32_bits(cmd_buffer->paddr), tee->io_regs + tee->vdata->cmdbuff_addr_hi_reg); iowrite32(TEE_RING_INIT_CMD, tee->io_regs + tee->vdata->cmdresp_reg); @@ -157,7 +197,7 @@ static int tee_init_ring(struct psp_tee_device *tee) } free_buf: - tee_free_cmd_buffer(cmd); + tee_free_cmd_buffer(cmd_buffer); return ret; } @@ -167,7 +207,7 @@ static void tee_destroy_ring(struct psp_tee_device *tee) unsigned int reg; int ret; - if (!tee->rb_mgr.ring_start) + if (!tee->rb_mgr.ring_buf->vaddr) return; if (psp_dead) @@ -256,7 +296,7 @@ static int tee_submit_cmd(struct psp_tee_device *tee, enum tee_cmd_id cmd_id, do { /* Get pointer to ring buffer command entry */ cmd = (struct tee_ring_cmd *) - (tee->rb_mgr.ring_start + tee->rb_mgr.wptr); + (tee->rb_mgr.ring_buf->vaddr + tee->rb_mgr.wptr); rptr = ioread32(tee->io_regs + tee->vdata->ring_rptr_reg); @@ -305,7 +345,7 @@ static int tee_submit_cmd(struct psp_tee_device *tee, enum tee_cmd_id cmd_id, /* Update local copy of write pointer */ tee->rb_mgr.wptr += sizeof(struct tee_ring_cmd); - if (tee->rb_mgr.wptr >= tee->rb_mgr.ring_size) + if (tee->rb_mgr.wptr >= tee->rb_mgr.ring_buf->size) tee->rb_mgr.wptr = 0; /* Trigger interrupt to Trusted OS */ diff --git a/drivers/crypto/ccp/tee-dev.h b/drivers/crypto/ccp/tee-dev.h index 49d26158b71e..ee22d60177a2 100644 --- a/drivers/crypto/ccp/tee-dev.h +++ b/drivers/crypto/ccp/tee-dev.h @@ -16,6 +16,7 @@ #include <linux/device.h> #include <linux/mutex.h> +#include <linux/psp-tee.h> #define TEE_DEFAULT_TIMEOUT 10 #define MAX_BUFFER_SIZE 988 @@ -48,16 +49,12 @@ struct tee_init_ring_cmd { /** * struct ring_buf_manager - Helper structure to manage ring buffer. - * @ring_start: starting address of ring buffer - * @ring_size: size of ring buffer in bytes - * @ring_pa: physical address of ring buffer + * @ring_buf: ring buffer allocated for sharing with TEE * @wptr: index to the last written entry in ring buffer */ struct ring_buf_manager { struct mutex mutex; /* synchronizes access to ring buffer */ - void *ring_start; - u32 ring_size; - phys_addr_t ring_pa; + struct psp_tee_buffer *ring_buf; u32 wptr; }; diff --git a/include/linux/psp-tee.h b/include/linux/psp-tee.h index cb0c95d6d76b..670bb7877456 100644 --- a/include/linux/psp-tee.h +++ b/include/linux/psp-tee.h @@ -13,6 +13,7 @@ #include <linux/types.h> #include <linux/errno.h> +#include <linux/dma-mapping.h> /* This file defines the Trusted Execution Environment (TEE) interface commands * and the API exported by AMD Secure Processor driver to communicate with @@ -40,6 +41,22 @@ enum tee_cmd_id { TEE_CMD_ID_UNMAP_SHARED_MEM, }; +/** + * struct psp_tee_buffer - Structure that has buffer details of memory that can + * be shared with PSP TEE. + * + * @dma: DMA address of memory + * @paddr: Physical address of memory + * @vaddr: CPU virtual address of memory + * @size: Size of memory in bytes + */ +struct psp_tee_buffer { + dma_addr_t dma; + phys_addr_t paddr; + void *vaddr; + unsigned long size; +}; + #ifdef CONFIG_CRYPTO_DEV_SP_PSP /** * psp_tee_process_cmd() - Process command in Trusted Execution Environment @@ -75,6 +92,28 @@ int psp_tee_process_cmd(enum tee_cmd_id cmd_id, void *buf, size_t len, */ int psp_check_tee_status(void); +/** + * psp_tee_alloc_buffer() - Allocates memory of requested size using + * dma_alloc_coherent() API. + * + * This function can be used to allocate a shared memory region between the + * host and PSP TEE. + * + * Returns: + * non-NULL a valid pointer to struct psp_tee_buffer + * NULL on failure + */ +struct psp_tee_buffer *psp_tee_alloc_buffer(unsigned long size); + +/** + * psp_tee_free_buffer() - Deallocates memory using dma_free_coherent() API. + * + * This function can be used to release shared memory region between host + * and PSP TEE. + * + */ +void psp_tee_free_buffer(struct psp_tee_buffer *buffer); + #else /* !CONFIG_CRYPTO_DEV_SP_PSP */ static inline int psp_tee_process_cmd(enum tee_cmd_id cmd_id, void *buf, @@ -87,5 +126,15 @@ static inline int psp_check_tee_status(void) { return -ENODEV; } + +static inline +struct psp_tee_buffer *psp_tee_alloc_buffer(unsigned long size) +{ + return NULL; +} + +static inline void psp_tee_free_buffer(struct psp_tee_buffer *buffer) +{ +} #endif /* CONFIG_CRYPTO_DEV_SP_PSP */ #endif /* __PSP_TEE_H_ */ -- 2.25.1

2 years, 10 months

[PATCH AUTOSEL 6.0 7/7] drm/amdkfd: Fix double release compute pasid

by Sasha Levin

From: Philip Yang <Philip.Yang(a)amd.com> [ Upstream commit 1a799c4c190ea9f0e81028e3eb3037ed0ab17ff5 ] If kfd_process_device_init_vm returns failure after vm is converted to compute vm and vm->pasid set to compute pasid, KFD will not take pdd->drm_file reference. As a result, drm close file handler maybe called to release the compute pasid before KFD process destroy worker to release the same pasid and set vm->pasid to zero, this generates below WARNING backtrace and NULL pointer access. Add helper amdgpu_amdkfd_gpuvm_set_vm_pasid and call it at the last step of kfd_process_device_init_vm, to ensure vm pasid is the original pasid if acquiring vm failed or is the compute pasid with pdd->drm_file reference taken to avoid double release same pasid. amdgpu: Failed to create process VM object ida_free called for id=32770 which is not allocated. WARNING: CPU: 57 PID: 72542 at ../lib/idr.c:522 ida_free+0x96/0x140 RIP: 0010:ida_free+0x96/0x140 Call Trace: amdgpu_pasid_free_delayed+0xe1/0x2a0 [amdgpu] amdgpu_driver_postclose_kms+0x2d8/0x340 [amdgpu] drm_file_free.part.13+0x216/0x270 [drm] drm_close_helper.isra.14+0x60/0x70 [drm] drm_release+0x6e/0xf0 [drm] __fput+0xcc/0x280 ____fput+0xe/0x20 task_work_run+0x96/0xc0 do_exit+0x3d0/0xc10 BUG: kernel NULL pointer dereference, address: 0000000000000000 RIP: 0010:ida_free+0x76/0x140 Call Trace: amdgpu_pasid_free_delayed+0xe1/0x2a0 [amdgpu] amdgpu_driver_postclose_kms+0x2d8/0x340 [amdgpu] drm_file_free.part.13+0x216/0x270 [drm] drm_close_helper.isra.14+0x60/0x70 [drm] drm_release+0x6e/0xf0 [drm] __fput+0xcc/0x280 ____fput+0xe/0x20 task_work_run+0x96/0xc0 do_exit+0x3d0/0xc10 Signed-off-by: Philip Yang <Philip.Yang(a)amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling(a)amd.com> Signed-off-by: Alex Deucher <alexander.deucher(a)amd.com> Signed-off-by: Sasha Levin <sashal(a)kernel.org> --- drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h | 4 +- .../gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c | 39 +++++++++++++------ drivers/gpu/drm/amd/amdkfd/kfd_process.c | 12 ++++-- 3 files changed, 40 insertions(+), 15 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h index 647220a8762d..30f145dc8724 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h @@ -265,8 +265,10 @@ int amdgpu_amdkfd_get_pcie_bandwidth_mbytes(struct amdgpu_device *adev, bool is_ (&((struct amdgpu_fpriv *) \ ((struct drm_file *)(drm_priv))->driver_priv)->vm) +int amdgpu_amdkfd_gpuvm_set_vm_pasid(struct amdgpu_device *adev, + struct file *filp, u32 pasid); int amdgpu_amdkfd_gpuvm_acquire_process_vm(struct amdgpu_device *adev, - struct file *filp, u32 pasid, + struct file *filp, void **process_info, struct dma_fence **ef); void amdgpu_amdkfd_gpuvm_release_process_vm(struct amdgpu_device *adev, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c index 5e184952ec98..f86a132bb761 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c @@ -1471,10 +1471,9 @@ static void amdgpu_amdkfd_gpuvm_unpin_bo(struct amdgpu_bo *bo) amdgpu_bo_unreserve(bo); } -int amdgpu_amdkfd_gpuvm_acquire_process_vm(struct amdgpu_device *adev, - struct file *filp, u32 pasid, - void **process_info, - struct dma_fence **ef) +int amdgpu_amdkfd_gpuvm_set_vm_pasid(struct amdgpu_device *adev, + struct file *filp, u32 pasid) + { struct amdgpu_fpriv *drv_priv; struct amdgpu_vm *avm; @@ -1485,10 +1484,6 @@ int amdgpu_amdkfd_gpuvm_acquire_process_vm(struct amdgpu_device *adev, return ret; avm = &drv_priv->vm; - /* Already a compute VM? */ - if (avm->process_info) - return -EINVAL; - /* Free the original amdgpu allocated pasid, * will be replaced with kfd allocated pasid. */ @@ -1497,14 +1492,36 @@ int amdgpu_amdkfd_gpuvm_acquire_process_vm(struct amdgpu_device *adev, amdgpu_vm_set_pasid(adev, avm, 0); } - /* Convert VM into a compute VM */ - ret = amdgpu_vm_make_compute(adev, avm); + ret = amdgpu_vm_set_pasid(adev, avm, pasid); if (ret) return ret; - ret = amdgpu_vm_set_pasid(adev, avm, pasid); + return 0; +} + +int amdgpu_amdkfd_gpuvm_acquire_process_vm(struct amdgpu_device *adev, + struct file *filp, + void **process_info, + struct dma_fence **ef) +{ + struct amdgpu_fpriv *drv_priv; + struct amdgpu_vm *avm; + int ret; + + ret = amdgpu_file_to_fpriv(filp, &drv_priv); if (ret) return ret; + avm = &drv_priv->vm; + + /* Already a compute VM? */ + if (avm->process_info) + return -EINVAL; + + /* Convert VM into a compute VM */ + ret = amdgpu_vm_make_compute(adev, avm); + if (ret) + return ret; + /* Initialize KFD part of the VM and process info */ ret = init_kfd_vm(avm, process_info, ef); if (ret) diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c index 04678f9e214b..febf0e9f7af1 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c @@ -1581,9 +1581,9 @@ int kfd_process_device_init_vm(struct kfd_process_device *pdd, p = pdd->process; dev = pdd->dev; - ret = amdgpu_amdkfd_gpuvm_acquire_process_vm( - dev->adev, drm_file, p->pasid, - &p->kgd_process_info, &p->ef); + ret = amdgpu_amdkfd_gpuvm_acquire_process_vm(dev->adev, drm_file, + &p->kgd_process_info, + &p->ef); if (ret) { pr_err("Failed to create process VM object\n"); return ret; @@ -1598,10 +1598,16 @@ int kfd_process_device_init_vm(struct kfd_process_device *pdd, if (ret) goto err_init_cwsr; + ret = amdgpu_amdkfd_gpuvm_set_vm_pasid(dev->adev, drm_file, p->pasid); + if (ret) + goto err_set_pasid; + pdd->drm_file = drm_file; return 0; +err_set_pasid: + kfd_process_device_destroy_cwsr_dgpu(pdd); err_init_cwsr: kfd_process_device_destroy_ib_mem(pdd); err_reserve_ib_mem: -- 2.35.1

2 years, 10 months

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

Linaro-mm-sig