After walking and trying to clean up the worst in the driver, I came across the pci_driver::remove() _empty_ implementation. That would crash the system at least during hot-unplug (or write to remove in sysfs). There are many other problems: * Initialization + deinitialization apparently comes from no-hotplug support age. It needs a rewrite. * Hairy debug macros. Drop them. * Use of self-baked lists. Replace by list. * The order of the functions should be inverted and fwd decls dropped. * Coding style from the stone age. Fix. * I assume there are many bugs, but the code is unreadable at times, so hard to judge. There is one example posted [1]. I was able to find only one user back in 2016. So mark the driver as BROKEN for some time. Either someone will notice and we can bring the driver to this century. Or we will drop it completely if noone cares. [1] https://lore.kernel.org/all/20230728123901.64225-1-dg573847474@xxxxxxxxx/ Signed-off-by: Jiri Slaby (SUSE) <jirislaby@xxxxxxxxxx> Cc: Chengfeng Ye <dg573847474@xxxxxxxxx> --- drivers/tty/Kconfig | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/tty/Kconfig b/drivers/tty/Kconfig index 341abaed4ce2..907a7cb1d186 100644 --- a/drivers/tty/Kconfig +++ b/drivers/tty/Kconfig @@ -236,6 +236,7 @@ config MOXA_SMARTIO config SYNCLINK_GT tristate "SyncLink GT/AC support" depends on SERIAL_NONSTANDARD && PCI + depends on BROKEN help Support for SyncLink GT and SyncLink AC families of synchronous and asynchronous serial adapters -- 2.41.0
