Commit 403249c7 authored by Akiva Goldberger's avatar Akiva Goldberger Committed by Zhengchao Shao
Browse files

net/mlx5: Add a timeout to acquire the command queue semaphore

stable inclusion
from stable-v6.6.33
commit f9caccdd42e999b74303c9b0643300073ed5d319
category: bugfix
bugzilla: https://gitee.com/src-openeuler/kernel/issues/IA6S89
CVE: CVE-2024-38556

Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=f9caccdd42e999b74303c9b0643300073ed5d319



---------------------------

[ Upstream commit 485d65e1357123a697c591a5aeb773994b247ad7 ]

Prevent forced completion handling on an entry that has not yet been
assigned an index, causing an out of bounds access on idx = -22.
Instead of waiting indefinitely for the sem, blocking flow now waits for
index to be allocated or a sem acquisition timeout before beginning the
timer for FW completion.

Kernel log example:
mlx5_core 0000:06:00.0: wait_func_handle_exec_timeout:1128:(pid 185911): cmd[-22]: CREATE_UCTX(0xa04) No done completion

Fixes: 8e715cd6 ("net/mlx5: Set command entry semaphore up once got index free")
Signed-off-by: default avatarAkiva Goldberger <agoldberger@nvidia.com>
Reviewed-by: default avatarMoshe Shemesh <moshe@nvidia.com>
Signed-off-by: default avatarTariq Toukan <tariqt@nvidia.com>
Link: https://lore.kernel.org/r/20240509112951.590184-5-tariqt@nvidia.com


Signed-off-by: default avatarJakub Kicinski <kuba@kernel.org>
Signed-off-by: default avatarSasha Levin <sashal@kernel.org>
Signed-off-by: default avatarZhengchao Shao <shaozhengchao@huawei.com>
parent 26e63d23
Loading
Loading
Loading
Loading
+32 −9
Original line number Diff line number Diff line
@@ -967,19 +967,32 @@ static void cmd_work_handler(struct work_struct *work)
	bool poll_cmd = ent->polling;
	struct mlx5_cmd_layout *lay;
	struct mlx5_core_dev *dev;
	unsigned long cb_timeout;
	struct semaphore *sem;
	unsigned long timeout;
	unsigned long flags;
	int alloc_ret;
	int cmd_mode;

	complete(&ent->handling);

	dev = container_of(cmd, struct mlx5_core_dev, cmd);
	cb_timeout = msecs_to_jiffies(mlx5_tout_ms(dev, CMD));
	timeout = msecs_to_jiffies(mlx5_tout_ms(dev, CMD));

	complete(&ent->handling);
	sem = ent->page_queue ? &cmd->vars.pages_sem : &cmd->vars.sem;
	down(sem);
	if (!ent->page_queue) {
		if (down_timeout(&cmd->vars.sem, timeout)) {
			mlx5_core_warn(dev, "%s(0x%x) timed out while waiting for a slot.\n",
				       mlx5_command_str(ent->op), ent->op);
			if (ent->callback) {
				ent->callback(-EBUSY, ent->context);
				mlx5_free_cmd_msg(dev, ent->out);
				free_msg(dev, ent->in);
				cmd_ent_put(ent);
			} else {
				ent->ret = -EBUSY;
				complete(&ent->done);
			}
			complete(&ent->slotted);
			return;
		}
		alloc_ret = cmd_alloc_index(cmd, ent);
		if (alloc_ret < 0) {
			mlx5_core_err_rl(dev, "failed to allocate command entry\n");
@@ -992,10 +1005,11 @@ static void cmd_work_handler(struct work_struct *work)
				ent->ret = -EAGAIN;
				complete(&ent->done);
			}
			up(sem);
			up(&cmd->vars.sem);
			return;
		}
	} else {
		down(&cmd->vars.pages_sem);
		ent->idx = cmd->vars.max_reg_cmds;
		spin_lock_irqsave(&cmd->alloc_lock, flags);
		clear_bit(ent->idx, &cmd->vars.bitmask);
@@ -1003,6 +1017,8 @@ static void cmd_work_handler(struct work_struct *work)
		spin_unlock_irqrestore(&cmd->alloc_lock, flags);
	}

	complete(&ent->slotted);

	lay = get_inst(cmd, ent->idx);
	ent->lay = lay;
	memset(lay, 0, sizeof(*lay));
@@ -1021,7 +1037,7 @@ static void cmd_work_handler(struct work_struct *work)
	ent->ts1 = ktime_get_ns();
	cmd_mode = cmd->mode;

	if (ent->callback && schedule_delayed_work(&ent->cb_timeout_work, cb_timeout))
	if (ent->callback && schedule_delayed_work(&ent->cb_timeout_work, timeout))
		cmd_ent_get(ent);
	set_bit(MLX5_CMD_ENT_STATE_PENDING_COMP, &ent->state);

@@ -1141,6 +1157,9 @@ static int wait_func(struct mlx5_core_dev *dev, struct mlx5_cmd_work_ent *ent)
		ent->ret = -ECANCELED;
		goto out_err;
	}

	wait_for_completion(&ent->slotted);

	if (cmd->mode == CMD_MODE_POLLING || ent->polling)
		wait_for_completion(&ent->done);
	else if (!wait_for_completion_timeout(&ent->done, timeout))
@@ -1155,6 +1174,9 @@ static int wait_func(struct mlx5_core_dev *dev, struct mlx5_cmd_work_ent *ent)
	} else if (err == -ECANCELED) {
		mlx5_core_warn(dev, "%s(0x%x) canceled on out of queue timeout.\n",
			       mlx5_command_str(ent->op), ent->op);
	} else if (err == -EBUSY) {
		mlx5_core_warn(dev, "%s(0x%x) timeout while waiting for command semaphore.\n",
			       mlx5_command_str(ent->op), ent->op);
	}
	mlx5_core_dbg(dev, "err %d, delivery status %s(%d)\n",
		      err, deliv_status_to_str(ent->status), ent->status);
@@ -1206,6 +1228,7 @@ static int mlx5_cmd_invoke(struct mlx5_core_dev *dev, struct mlx5_cmd_msg *in,
	ent->polling = force_polling;

	init_completion(&ent->handling);
	init_completion(&ent->slotted);
	if (!callback)
		init_completion(&ent->done);

@@ -1223,7 +1246,7 @@ static int mlx5_cmd_invoke(struct mlx5_core_dev *dev, struct mlx5_cmd_msg *in,
		return 0; /* mlx5_cmd_comp_handler() will put(ent) */

	err = wait_func(dev, ent);
	if (err == -ETIMEDOUT || err == -ECANCELED)
	if (err == -ETIMEDOUT || err == -ECANCELED || err == -EBUSY)
		goto out_free;

	ds = ent->ts2 - ent->ts1;
+1 −0
Original line number Diff line number Diff line
@@ -852,6 +852,7 @@ struct mlx5_cmd_work_ent {
	void		       *context;
	int			idx;
	struct completion	handling;
	struct completion	slotted;
	struct completion	done;
	struct mlx5_cmd        *cmd;
	struct work_struct	work;