On 2024/03/28 11:48, Guanyou Chen wrote: hmm, I don't have an arm64 machine, but its code doesn't looks so.. any history? https://github.com/torvalds/linux/blob/master/arch/arm64/kernel/vmcore_in... vmcoreinfo_append_str("NUMBER(TCR_EL1_T1SZ)=0x%llx\n", get_tcr_el1_t1sz()); Thanks, Kazu > > Before: > vmcoreinfo : vabits_actual: 27 > > After: > vmcoreinfo : vabits_actual: 39 > > Fixes: 28f0833037b2 ("Fix arm64: rewrite the arm64_get_vmcoreinfo_ul to > arm64_get_vmcoreinfo") > > Signed-off-by: chenguanyou <chenguanyou(a)xiaomi.com > <mailto:chenguanyou@xiaomi.com>> > --- >  arm64.c | 4 ++-- >  1 file changed, 2 insertions(+), 2 deletions(-) > > diff --git a/arm64.c b/arm64.c > index af0e0d7..e36c723 100644 > --- a/arm64.c > +++ b/arm64.c > @@ -4656,8 +4656,8 @@ arm64_set_va_bits_by_tcr(void) >  { >     ulong value; > > -   if (arm64_get_vmcoreinfo(&value, "NUMBER(TCR_EL1_T1SZ)", NUM_HEX) || > -       arm64_get_vmcoreinfo(&value, "NUMBER(tcr_el1_t1sz)", NUM_HEX)) { > +   if (arm64_get_vmcoreinfo(&value, "NUMBER(TCR_EL1_T1SZ)", NUM_DEC) || > +       arm64_get_vmcoreinfo(&value, "NUMBER(tcr_el1_t1sz)", NUM_DEC)) { >         /* See ARMv8 ARM for the description of >          * TCR_EL1.T1SZ and how it can be used >          * to calculate the vabits_actual > -- > 2.39.0 > > Thanks, > Guanyou

Hi Kazu You are right, The root cause is mtk vmcore build issue. ` "NUMBER(TCR_EL1_T1SZ)=%llu\n" ` in our code. Guanyou Chen <chenguanyou9338(a)gmail.com&gt; 于2024年3月28日周四 10:48写道：