drm/xe: don't allocate under ct->lock

Seems to be a sensitive lock, where ct->lock looks to be primed with
fs_reclaim, so holding that and then allocating memory will cause
lockdep to complain. We need to change the ordering wrt to grabbing the
ct->lock and potentially grabbing the runtime_pm, since some of the
runtime_pm routines can allocate memory (or at least that's what lockdep
seems to suggest).

Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

diff --git a/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c b/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c
index c815a42e2cdba18bbd416b92f263497767c4d77d..20f8f0aae6b42234d69fd2f2edc3278cb2dc0014 100644 (file)
--- a/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c
+++ b/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c
@@ -5,6 +5,7 @@
 
 #include "xe_gt_tlb_invalidation.h"
 
+#include "xe_device.h"
 #include "xe_gt.h"
 #include "xe_guc.h"
 #include "xe_guc_ct.h"
@@ -112,6 +113,8 @@ static int send_tlb_invalidation(struct xe_guc *guc,
         * in order which they currently are, if that changes the algorithm will
         * need to be updated.
         */
+
+       xe_device_mem_access_get(gt->xe);
        mutex_lock(&guc->ct.lock);
        seqno = gt->tlb_invalidation.seqno;
        if (fence) {
@@ -140,6 +143,7 @@ static int send_tlb_invalidation(struct xe_guc *guc,
        if (ret < 0 && fence)
                invalidation_fence_signal(fence);
        mutex_unlock(&guc->ct.lock);
+       xe_device_mem_access_put(gt->xe);
 
        return ret;
 }

diff --git a/drivers/gpu/drm/xe/xe_guc_ct.c b/drivers/gpu/drm/xe/xe_guc_ct.c
index e8c2edb1359d9b5fd58f65e4f8c22abbf049ebdf..9dc906f2651a55174de8a446c973fb91bcfa54de 100644 (file)
--- a/drivers/gpu/drm/xe/xe_guc_ct.c
+++ b/drivers/gpu/drm/xe/xe_guc_ct.c
@@ -498,26 +498,22 @@ static int __guc_ct_send_locked(struct xe_guc_ct *ct, const u32 *action,
                }
        }
 
-       xe_device_mem_access_get(ct_to_xe(ct));
 retry:
        ret = has_room(ct, len + GUC_CTB_HDR_LEN, g2h_len);
        if (unlikely(ret))
-               goto put_wa;
+               goto out;
 
        ret = h2g_write(ct, action, len, g2h_fence ? g2h_fence->seqno : 0,
                        !!g2h_fence);
        if (unlikely(ret)) {
                if (ret == -EAGAIN)
                        goto retry;
-               goto put_wa;
+               goto out;
        }
 
        g2h_reserve_space(ct, g2h_len, num_g2h);
        xe_guc_notify(ct_to_guc(ct));
-put_wa:
-       xe_device_mem_access_put(ct_to_xe(ct));
 out:
-
        return ret;
 }
 
@@ -539,6 +535,7 @@ static int guc_ct_send_locked(struct xe_guc_ct *ct, const u32 *action, u32 len,
 
        XE_BUG_ON(g2h_len && g2h_fence);
        lockdep_assert_held(&ct->lock);
+       xe_device_assert_mem_access(ct_to_xe(ct));
 
 try_again:
        ret = __guc_ct_send_locked(ct, action, len, g2h_len, num_g2h,
@@ -608,10 +605,14 @@ static int guc_ct_send(struct xe_guc_ct *ct, const u32 *action, u32 len,
 
        XE_BUG_ON(g2h_len && g2h_fence);
 
+       xe_device_mem_access_get(ct_to_xe(ct));
+
        mutex_lock(&ct->lock);
        ret = guc_ct_send_locked(ct, action, len, g2h_len, num_g2h, g2h_fence);
        mutex_unlock(&ct->lock);
 
+       xe_device_mem_access_put(ct_to_xe(ct));
+
        return ret;
 }