Re: [PATCH 5.10 1/2] iwlwifi: Fix softirq/hardirq disabling in iwl_pcie_gen2_enqueue_hcmd()

[Jari Ruusu <jariruusu@protonmail.com>]

On Tuesday, May 4, 2021 6:35 PM, Greg Kroah-Hartman <gregkh@linuxfoundation.org> wrote:
> Can you resend your backports here now, and properly let us know what
> kernel they belong into (again, the subject line is very confusing.)

iwlwifi: Fix softirq/hardirq disabling in iwl_pcie_gen2_enqueue_hcmd()
upstream commit e7020bb068d8be50a92f48e36b236a1a1ef9282e,
backport for linux-5.4.y and linux-4.19.y (booted and ping tested)
Signed-off-by: Jari Ruusu <jariruusu@protonmail.com>

--- a/drivers/net/wireless/intel/iwlwifi/pcie/tx-gen2.c
+++ b/drivers/net/wireless/intel/iwlwifi/pcie/tx-gen2.c
@@ -705,6 +705,7 @@ static int iwl_pcie_gen2_enqueue_hcmd(struct iwl_trans *trans,
 	const u8 *cmddata[IWL_MAX_CMD_TBS_PER_TFD];
 	u16 cmdlen[IWL_MAX_CMD_TBS_PER_TFD];
 	struct iwl_tfh_tfd *tfd;
+	unsigned long flags2;

 	copy_size = sizeof(struct iwl_cmd_header_wide);
 	cmd_size = sizeof(struct iwl_cmd_header_wide);
@@ -773,14 +774,14 @@ static int iwl_pcie_gen2_enqueue_hcmd(struct iwl_trans *trans,
 		goto free_dup_buf;
 	}

-	spin_lock_bh(&txq->lock);
+	spin_lock_irqsave(&txq->lock, flags2);

 	idx = iwl_pcie_get_cmd_index(txq, txq->write_ptr);
 	tfd = iwl_pcie_get_tfd(trans, txq, txq->write_ptr);
 	memset(tfd, 0, sizeof(*tfd));

 	if (iwl_queue_space(trans, txq) < ((cmd->flags & CMD_ASYNC) ? 2 : 1)) {
-		spin_unlock_bh(&txq->lock);
+		spin_unlock_irqrestore(&txq->lock, flags2);

 		IWL_ERR(trans, "No space in command queue\n");
 		iwl_op_mode_cmd_queue_full(trans->op_mode);
@@ -915,7 +916,7 @@ static int iwl_pcie_gen2_enqueue_hcmd(struct iwl_trans *trans,
 	spin_unlock_irqrestore(&trans_pcie->reg_lock, flags);

 out:
-	spin_unlock_bh(&txq->lock);
+	spin_unlock_irqrestore(&txq->lock, flags2);
 free_dup_buf:
 	if (idx < 0)
 		kfree(dup_buf);

--
Jari Ruusu  4096R/8132F189 12D6 4C3A DCDA 0AA4 27BD  ACDF F073 3C80 8132 F189