Re: [RFC PATCH net-next v4 05/11] net/smc: implement DMB-related operations of loopback-ism

Wen Gu <guwen@xxxxxxxxxxxxxxxxx> · Fri, 22 Mar 2024 20:30:03 +0800

On 2024/3/21 16:12, Jan Karcher wrote:


On 17/03/2024 11:05, Wen Gu wrote:
This implements DMB (un)registration and data move operations of
loopback-ism device.

Signed-off-by: Wen Gu <guwen@xxxxxxxxxxxxxxxxx>
---
  net/smc/smc_loopback.c | 131 ++++++++++++++++++++++++++++++++++++++++-
  net/smc/smc_loopback.h |  13 ++++
  2 files changed, 141 insertions(+), 3 deletions(-)

diff --git a/net/smc/smc_loopback.c b/net/smc/smc_loopback.c
index 253128c77208..7335acb03920 100644
--- a/net/smc/smc_loopback.c
+++ b/net/smc/smc_loopback.c
@@ -15,11 +15,13 @@
  #include <linux/types.h>
  #include <net/smc.h>
+#include "smc_cdc.h"
  #include "smc_ism.h"
  #include "smc_loopback.h"
  #if IS_ENABLED(CONFIG_SMC_LO)
  #define SMC_LO_V2_CAPABLE    0x1 /* loopback-ism acts as ISMv2 */
+#define SMC_DMA_ADDR_INVALID    (~(dma_addr_t)0)
  static const char smc_lo_dev_name[] = "loopback-ism";
  static struct smc_lo_dev *lo_dev;
@@ -49,6 +51,93 @@ static int smc_lo_query_rgid(struct smcd_dev *smcd, struct smcd_gid *rgid,
      return 0;
  }
+static int smc_lo_register_dmb(struct smcd_dev *smcd, struct smcd_dmb *dmb,
+                   void *client_priv)
+{
+    struct smc_lo_dmb_node *dmb_node, *tmp_node;
+    struct smc_lo_dev *ldev = smcd->priv;
+    int sba_idx, rc;
+
+    /* check space for new dmb */
+    for_each_clear_bit(sba_idx, ldev->sba_idx_mask, SMC_LO_MAX_DMBS) {
+        if (!test_and_set_bit(sba_idx, ldev->sba_idx_mask))
+            break;
+    }
+    if (sba_idx == SMC_LO_MAX_DMBS)
+        return -ENOSPC;
+
+    dmb_node = kzalloc(sizeof(*dmb_node), GFP_KERNEL);
+    if (!dmb_node) {
+        rc = -ENOMEM;
+        goto err_bit;
+    }
+
+    dmb_node->sba_idx = sba_idx;
+    dmb_node->len = dmb->dmb_len;
+    dmb_node->cpu_addr = kzalloc(dmb_node->len, GFP_KERNEL |
+                     __GFP_NOWARN | __GFP_NORETRY |
+                     __GFP_NOMEMALLOC);
+    if (!dmb_node->cpu_addr) {
+        rc = -ENOMEM;
+        goto err_node;
+    }
+    dmb_node->dma_addr = SMC_DMA_ADDR_INVALID;
+
+again:
+    /* add new dmb into hash table */
+    get_random_bytes(&dmb_node->token, sizeof(dmb_node->token));
+    write_lock_bh(&ldev->dmb_ht_lock);
+    hash_for_each_possible(ldev->dmb_ht, tmp_node, list, dmb_node->token) {
+        if (tmp_node->token == dmb_node->token) {
+            write_unlock_bh(&ldev->dmb_ht_lock);
+            goto again;
+        }
+    }
+    hash_add(ldev->dmb_ht, &dmb_node->list, dmb_node->token);
+    write_unlock_bh(&ldev->dmb_ht_lock);
+
+    dmb->sba_idx = dmb_node->sba_idx;
+    dmb->dmb_tok = dmb_node->token;
+    dmb->cpu_addr = dmb_node->cpu_addr;
+    dmb->dma_addr = dmb_node->dma_addr;
+    dmb->dmb_len = dmb_node->len;
+
+    return 0;
+
+err_node:
+    kfree(dmb_node);
+err_bit:
+    clear_bit(sba_idx, ldev->sba_idx_mask);
+    return rc;
+}
+
+static int smc_lo_unregister_dmb(struct smcd_dev *smcd, struct smcd_dmb *dmb)
+{
+    struct smc_lo_dmb_node *dmb_node = NULL, *tmp_node;
+    struct smc_lo_dev *ldev = smcd->priv;
+
+    /* remove dmb from hash table */
+    write_lock_bh(&ldev->dmb_ht_lock);
+    hash_for_each_possible(ldev->dmb_ht, tmp_node, list, dmb->dmb_tok) {
+        if (tmp_node->token == dmb->dmb_tok) {
+            dmb_node = tmp_node;
+            break;
+        }
+    }
+    if (!dmb_node) {
+        write_unlock_bh(&ldev->dmb_ht_lock);
+        return -EINVAL;
+    }
+    hash_del(&dmb_node->list);
+    write_unlock_bh(&ldev->dmb_ht_lock);
+
+    clear_bit(dmb_node->sba_idx, ldev->sba_idx_mask);
+    kfree(dmb_node->cpu_addr);
+    kfree(dmb_node);
+
+    return 0;
+}
+
  static int smc_lo_add_vlan_id(struct smcd_dev *smcd, u64 vlan_id)
  {
      return -EOPNOTSUPP;
@@ -75,6 +164,40 @@ static int smc_lo_signal_event(struct smcd_dev *dev, struct smcd_gid *rgid,
      return 0;
  }
+static int smc_lo_move_data(struct smcd_dev *smcd, u64 dmb_tok,
+                unsigned int idx, bool sf, unsigned int offset,
+                void *data, unsigned int size)
+{
+    struct smc_lo_dmb_node *rmb_node = NULL, *tmp_node;
+    struct smc_lo_dev *ldev = smcd->priv;
+
+    read_lock_bh(&ldev->dmb_ht_lock);
+    hash_for_each_possible(ldev->dmb_ht, tmp_node, list, dmb_tok) {
+        if (tmp_node->token == dmb_tok) {
+            rmb_node = tmp_node;
+            break;
+        }
+    }
+    if (!rmb_node) {
+        read_unlock_bh(&ldev->dmb_ht_lock);
+        return -EINVAL;
+    }
+    read_unlock_bh(&ldev->dmb_ht_lock);
+
+    memcpy((char *)rmb_node->cpu_addr + offset, data, size);

Hi Wen Gu,

Could we get into use after free trouble here if the dmb gets unregistered between the read_unlock and memcpy?


rmb_node won't be unregistered until smc_lgr_free_bufs() in __smc_lgr_free(). At
that time, the connections on this lgr should be all freed (smc_conn_free() and
then lgr->refcnt == 0), so I think there will be no move data operation at that
point. But in case there is something unforeseen, I will put memcpy between dmb_ht_lock.

Thanks!


+
+    if (sf) {
+        struct smc_connection *conn =
+            smcd->conn[rmb_node->sba_idx];

Please put the `struct smc_connection *conn = NULL` at the top of the function and assign the value here.


OK, I will put it at the top. Thanks!

Thanks
- Jan

+
+        if (conn && !conn->killed)
+            tasklet_schedule(&conn->rx_tsklet);
+        else
+            return -EPIPE;
+    }
+    return 0;
+}
+
  static int smc_lo_supports_v2(void)
  {
      return SMC_LO_V2_CAPABLE;
@@ -101,14 +224,14 @@ static struct device *smc_lo_get_dev(struct smcd_dev *smcd)
  static const struct smcd_ops lo_ops = {
      .query_remote_gid = smc_lo_query_rgid,
-    .register_dmb        = NULL,
-    .unregister_dmb        = NULL,
+    .register_dmb = smc_lo_register_dmb,
+    .unregister_dmb = smc_lo_unregister_dmb,
      .add_vlan_id = smc_lo_add_vlan_id,
      .del_vlan_id = smc_lo_del_vlan_id,
      .set_vlan_required = smc_lo_set_vlan_required,
      .reset_vlan_required = smc_lo_reset_vlan_required,
      .signal_event = smc_lo_signal_event,
-    .move_data        = NULL,
+    .move_data = smc_lo_move_data,
      .supports_v2 = smc_lo_supports_v2,
      .get_local_gid = smc_lo_get_local_gid,
      .get_chid = smc_lo_get_chid,
@@ -173,6 +296,8 @@ static void smcd_lo_unregister_dev(struct smc_lo_dev *ldev)
  static int smc_lo_dev_init(struct smc_lo_dev *ldev)
  {
      smc_lo_generate_id(ldev);
+    rwlock_init(&ldev->dmb_ht_lock);
+    hash_init(ldev->dmb_ht);
      return smcd_lo_register_dev(ldev);
  }
diff --git a/net/smc/smc_loopback.h b/net/smc/smc_loopback.h
index 55b41133a97f..24ab9d747613 100644
--- a/net/smc/smc_loopback.h
+++ b/net/smc/smc_loopback.h
@@ -20,13 +20,26 @@
  #if IS_ENABLED(CONFIG_SMC_LO)
  #define SMC_LO_MAX_DMBS        5000
+#define SMC_LO_DMBS_HASH_BITS    12
  #define SMC_LO_CHID        0xFFFF
+struct smc_lo_dmb_node {
+    struct hlist_node list;
+    u64 token;
+    u32 len;
+    u32 sba_idx;
+    void *cpu_addr;
+    dma_addr_t dma_addr;
+};
+
  struct smc_lo_dev {
      struct smcd_dev *smcd;
      struct device dev;
      u16 chid;
      struct smcd_gid local_gid;
+    rwlock_t dmb_ht_lock;
+    DECLARE_BITMAP(sba_idx_mask, SMC_LO_MAX_DMBS);
+    DECLARE_HASHTABLE(dmb_ht, SMC_LO_DMBS_HASH_BITS);
  };
  #endif







