habanalabs: prevent CPU soft lockup on Palladium

Unmapping ptes in the device MMU on Palladium can take a long time, which
can cause a kernel BUG of CPU soft lockup.

This patch minimize the chances for this bug by sleeping a little between
unmapping ptes.

Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>

diff --git a/drivers/misc/habanalabs/memory.c b/drivers/misc/habanalabs/memory.c
index 39788b1cf8d0536066d5338b367003d98d62958b..e9e1635455845e985a0c315f1162ffd0a762cf61 100644 (file)
--- a/drivers/misc/habanalabs/memory.c
+++ b/drivers/misc/habanalabs/memory.c
@@ -1046,10 +1046,17 @@ static int unmap_device_va(struct hl_ctx *ctx, u64 vaddr)
 
        mutex_lock(&ctx->mmu_lock);
 
-       for (i = 0 ; i < phys_pg_pack->npages ; i++, next_vaddr += page_size)
+       for (i = 0 ; i < phys_pg_pack->npages ; i++, next_vaddr += page_size) {
                if (hl_mmu_unmap(ctx, next_vaddr, page_size))
                        dev_warn_ratelimited(hdev->dev,
-                               "unmap failed for vaddr: 0x%llx\n", next_vaddr);
+                       "unmap failed for vaddr: 0x%llx\n", next_vaddr);
+
+               /* unmapping on Palladium can be really long, so avoid a CPU
+                * soft lockup bug by sleeping a little between unmapping pages
+                */
+               if (hdev->pldm)
+                       usleep_range(500, 1000);
+       }
 
        hdev->asic_funcs->mmu_invalidate_cache(hdev, true);