crypto: hisilicon/qm - obtain the mailbox configuration at one time

The malibox needs to be triggered by a 128bit atomic operation.
The reason is that the PF and VFs of the device share the mmio memory
of the mailbox, and the mutex cannot lock mailbox operations in
different functions, especially when passing through VFs to
virtual machines.

Currently, the write operation to the mailbox is already a 128-bit
atomic write. The read operation also needs to be modified to a
128-bit atomic read. Since there is no general 128-bit IO memory
access API in the current ARM64 architecture, and the stp and ldp
instructions do not guarantee atomic access to device memory, they
cannot be extracted as a general API. Therefore, the 128-bit atomic
read and write operations need to be implemented in the driver.

Signed-off-by: Weili Qian <qianweili@huawei.com>
Signed-off-by: Chenghai Huang <huangchenghai2@huawei.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
This commit is contained in:
Weili Qian
2026-01-17 18:18:05 +08:00
committed by Herbert Xu
parent fc8ae11b84
commit 3296992ffc
2 changed files with 89 additions and 49 deletions

View File

@@ -571,6 +571,7 @@ void hisi_qm_reset_done(struct pci_dev *pdev);
int hisi_qm_wait_mb_ready(struct hisi_qm *qm);
int hisi_qm_mb(struct hisi_qm *qm, u8 cmd, dma_addr_t dma_addr, u16 queue,
bool op);
int hisi_qm_mb_read(struct hisi_qm *qm, u64 *base, u8 cmd, u16 queue);
struct hisi_acc_sgl_pool;
struct hisi_acc_hw_sgl *hisi_acc_sg_buf_map_to_hw_sgl(struct device *dev,