tty: goldfish: use guard() for locks

Using guard()s is cleaner and safer.

goldfish_tty_probe() is omitted due to the crossing err_unmap
goto-label. Using scoped_guard() does not look that nice there. Perhaps
if someone refactored the locked part into a separate function...

Signed-off-by: Jiri Slaby (SUSE) <jirislaby@kernel.org>
Link: https://patch.msgid.link/20260703084717.176442-4-jirislaby@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This commit is contained in:
Jiri Slaby (SUSE) 2026-07-03 10:47:17 +02:00 committed by Greg Kroah-Hartman
parent e31bd02f19
commit e508a176d8

View File

@ -61,10 +61,10 @@ static inline void gf_write_addr(unsigned long addr, void __iomem *portl, void _
static void do_rw_io(struct goldfish_tty *qtty, unsigned long address,
size_t count, bool is_write)
{
unsigned long irq_flags;
void __iomem *base = qtty->base;
spin_lock_irqsave(&qtty->lock, irq_flags);
guard(spinlock_irqsave)(&qtty->lock);
gf_write_addr(address, base + GOLDFISH_TTY_REG_DATA_PTR,
base + GOLDFISH_TTY_REG_DATA_PTR_HIGH);
gf_iowrite32(count, base + GOLDFISH_TTY_REG_DATA_LEN);
@ -75,8 +75,6 @@ static void do_rw_io(struct goldfish_tty *qtty, unsigned long address,
else
gf_iowrite32(GOLDFISH_TTY_CMD_READ_BUFFER,
base + GOLDFISH_TTY_REG_CMD);
spin_unlock_irqrestore(&qtty->lock, irq_flags);
}
static void goldfish_tty_rw(struct goldfish_tty *qtty, unsigned long addr,
@ -418,7 +416,7 @@ static void goldfish_tty_remove(struct platform_device *pdev)
{
struct goldfish_tty *qtty = platform_get_drvdata(pdev);
mutex_lock(&goldfish_tty_lock);
guard(mutex)(&goldfish_tty_lock);
unregister_console(&qtty->console);
tty_unregister_device(goldfish_tty_driver, qtty->console.index);
@ -429,7 +427,6 @@ static void goldfish_tty_remove(struct platform_device *pdev)
goldfish_tty_current_line_count--;
if (goldfish_tty_current_line_count == 0)
goldfish_tty_delete_driver();
mutex_unlock(&goldfish_tty_lock);
}
#ifdef CONFIG_GOLDFISH_TTY_EARLY_CONSOLE