tmakin/ack-tegra
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

ANDROID: KVM: arm64: Use the hypervisor lock for FF-A transactions

Browse Source 
Synchronize the access to the hypervisor FF-A buffers using the
hypervisor buffers lock instead of the host lock. This is required to
prevent concurrent access to the hypervisor pair of buffers when
multiple VMs attempt to use them as part of an FF-A transaction.

Bug: 269285339
Bug: 278749606
Change-Id: Id7f79ffc8d2bc20be591251991cd66e307746102
Signed-off-by: Sebastian Ene <sebastianene@google.com>
This commit is contained in:
Sebastian Ene
2023-02-14 16:43:59 +00:00
parent 40551de8b0
commit f282cf2ef0
1 changed files with 16 additions and 21 deletions
Download Patch File Download Diff File Expand all files Collapse all files
arch/arm64/kvm/hyp/nvhe/ffa.c
+16 -21
View File
@@ -55,21 +55,21 @@ struct kvm_ffa_descriptor_buffer {
static struct kvm_ffa_descriptor_buffer ffa_desc_buf;
struct kvm_ffa_buffers {
hyp_spinlock_t lock;
void *tx;
void *rx;
};
/*
* Note that we don't currently lock these buffers explicitly, instead
* relying on the locking of the host FFA buffers as we only have one
* client.
* relying on the locking of the hyp FFA buffers.
*/
static struct kvm_ffa_buffers hyp_buffers;
static struct kvm_ffa_buffers host_buffers;
static u32 hyp_ffa_version;
static bool has_version_negotiated;
static hyp_spinlock_t version_lock;
static DEFINE_HYP_SPINLOCK(version_lock);
static DEFINE_HYP_SPINLOCK(kvm_ffa_hyp_lock);
static void ffa_to_smccc_error(struct arm_smccc_res *res, u64 ffa_errno)
{
@@ -208,7 +208,7 @@ static void do_ffa_rxtx_map(struct arm_smccc_res *res,
goto out;
}
hyp_spin_lock(&host_buffers.lock);
hyp_spin_lock(&kvm_ffa_hyp_lock);
if (host_buffers.tx) {
ret = FFA_RET_DENIED;
goto out_unlock;
@@ -252,7 +252,7 @@ static void do_ffa_rxtx_map(struct arm_smccc_res *res,
host_buffers.rx = rx_virt;
out_unlock:
hyp_spin_unlock(&host_buffers.lock);
hyp_spin_unlock(&kvm_ffa_hyp_lock);
out:
ffa_to_smccc_res(res, ret);
return;
@@ -280,7 +280,7 @@ static void do_ffa_rxtx_unmap(struct arm_smccc_res *res,
goto out;
}
hyp_spin_lock(&host_buffers.lock);
hyp_spin_lock(&kvm_ffa_hyp_lock);
if (!host_buffers.tx) {
ret = FFA_RET_INVALID_PARAMETERS;
goto out_unlock;
@@ -297,7 +297,7 @@ static void do_ffa_rxtx_unmap(struct arm_smccc_res *res,
ffa_unmap_hyp_buffers();
out_unlock:
hyp_spin_unlock(&host_buffers.lock);
hyp_spin_unlock(&kvm_ffa_hyp_lock);
out:
ffa_to_smccc_res(res, ret);
}
@@ -388,7 +388,7 @@ static void do_ffa_mem_frag_tx(struct arm_smccc_res *res,
if (fraglen % sizeof(*buf))
goto out;
hyp_spin_lock(&host_buffers.lock);
hyp_spin_lock(&kvm_ffa_hyp_lock);
if (!host_buffers.tx)
goto out_unlock;
@@ -413,7 +413,7 @@ static void do_ffa_mem_frag_tx(struct arm_smccc_res *res,
WARN_ON(ffa_host_unshare_ranges(buf, nr_ranges));
out_unlock:
hyp_spin_unlock(&host_buffers.lock);
hyp_spin_unlock(&kvm_ffa_hyp_lock);
out:
if (ret)
ffa_to_smccc_res(res, ret);
@@ -456,7 +456,7 @@ static void __do_ffa_mem_xfer(const u64 func_id,
goto out;
}
hyp_spin_lock(&host_buffers.lock);
hyp_spin_lock(&kvm_ffa_hyp_lock);
if (!host_buffers.tx) {
ret = FFA_RET_INVALID_PARAMETERS;
goto out_unlock;
@@ -507,7 +507,7 @@ static void __do_ffa_mem_xfer(const u64 func_id,
}
out_unlock:
hyp_spin_unlock(&host_buffers.lock);
hyp_spin_unlock(&kvm_ffa_hyp_lock);
out:
if (ret)
ffa_to_smccc_res(res, ret);
@@ -541,7 +541,7 @@ static void do_ffa_mem_reclaim(struct arm_smccc_res *res,
handle = PACK_HANDLE(handle_lo, handle_hi);
hyp_spin_lock(&host_buffers.lock);
hyp_spin_lock(&kvm_ffa_hyp_lock);
buf = hyp_buffers.tx;
*buf = (struct ffa_mem_region) {
@@ -603,7 +603,7 @@ static void do_ffa_mem_reclaim(struct arm_smccc_res *res,
WARN_ON(ffa_host_unshare_ranges(reg->constituents,
reg->addr_range_cnt));
out_unlock:
hyp_spin_unlock(&host_buffers.lock);
hyp_spin_unlock(&kvm_ffa_hyp_lock);
if (ret)
ffa_to_smccc_res(res, ret);
@@ -756,7 +756,7 @@ static void do_ffa_part_get(struct arm_smccc_res *res,
DECLARE_REG(u32, flags, ctxt, 5);
u32 count, partition_sz, copy_sz;
hyp_spin_lock(&host_buffers.lock);
hyp_spin_lock(&kvm_ffa_hyp_lock);
if (!host_buffers.rx) {
ffa_to_smccc_res(res, FFA_RET_BUSY);
goto out_unlock;
@@ -792,7 +792,7 @@ static void do_ffa_part_get(struct arm_smccc_res *res,
memcpy(host_buffers.rx, hyp_buffers.rx, copy_sz);
out_unlock:
hyp_spin_unlock(&host_buffers.lock);
hyp_spin_unlock(&kvm_ffa_hyp_lock);
}
bool kvm_host_ffa_handler(struct kvm_cpu_context *host_ctxt, u32 func_id)
@@ -908,15 +908,10 @@ int hyp_ffa_init(void *pages)
};
hyp_buffers = (struct kvm_ffa_buffers) {
.lock = __HYP_SPIN_LOCK_UNLOCKED,
.tx = tx,
.rx = rx,
};
host_buffers = (struct kvm_ffa_buffers) {
.lock = __HYP_SPIN_LOCK_UNLOCKED,
};
version_lock = __HYP_SPIN_LOCK_UNLOCKED;
return 0;
}