On Mon, Dec 30, 2024 at 02:24:03PM +0100, Borislav Petkov wrote:
> On Mon, Dec 30, 2024 at 09:04:11PM +0800, Feng Tang wrote:
> > As per kernel config, most ARCH has 'panic_timeout' as 0 by default, so
> > need to set the kcmdline. For the case in my commit log, where user had
> > clear requirement for not-reboot and wait, the manually set 'panic=0'
> > should take priority here?
>
> I think so.
>
> I'm not convinced that lets-log-the-hw-error-faster should override the panic
> timeout setting but I'm open to real-life example scenarios...
>
I see. Upon the discussion so far, how about the change below?
---
diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 07789f0b59bc..113471b76d8d 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -173,8 +173,6 @@ static struct gen_pool *ghes_estatus_pool;
static struct ghes_estatus_cache __rcu *ghes_estatus_caches[GHES_ESTATUS_CACHES_SIZE];
static atomic_t ghes_estatus_cache_alloced;
-static int ghes_panic_timeout __read_mostly = 30;
-
static void __iomem *ghes_map(u64 pfn, enum fixed_addresses fixmap_idx)
{
phys_addr_t paddr;
@@ -987,9 +985,10 @@ static void __ghes_panic(struct ghes *ghes,
ghes_clear_estatus(ghes, estatus, buf_paddr, fixmap_idx);
- /* reboot to log the error! */
- if (!panic_timeout)
- panic_timeout = ghes_panic_timeout;
+ /* If user hasn't specifically set panic timeout, reboot to log the error! */
+ if (!panic_timeout && !strstr(saved_command_line, "panic="))
+ panic_timeout = 30;
+
panic("Fatal hardware error!");
}
Or we want to stick the orignal patch, which doesn't change the
original flow?
Thanks,
Feng
