We're having an issue where we have a recent clang that seems
to generate kind_flag for enums (aka, adding signed/unsigned).
Trying to install a module on a kernel that doesn't have commit
6089fb325cf7 ("bpf: Add btf enum64 support") returns the following:
[ 3.176954] BPF:Invalid btf_info kind_flag
The enum that it complains about doesn't seem to have anything special
except having a sign:
[1721] ENUM 'perf_event_state' encoding=SIGNED size=4 vlen=6
'PERF_EVENT_STATE_DEAD' val=-4
'PERF_EVENT_STATE_EXIT' val=-3
'PERF_EVENT_STATE_ERROR' val=-2
'PERF_EVENT_STATE_OFF' val=-1
'PERF_EVENT_STATE_INACTIVE' val=0
'PERF_EVENT_STATE_ACTIVE' val=1
We are not currently using CONFIG_DEBUG_INFO_BTF_MODULES and
don't plan to use module BTF, so it's preferable to be able
to explicits disable it in the kernel config. Unfortunately,
because that kconfig option doesn't have a name, it's not
possible to flip it independently from CONFIG_DEBUG_INFO_BTF.
Let's add a name to make sure module BTF is user-controllable.
(Not sure, but maybe the right fix is to also have a stable patch
to relax that "Invalid btf_info kind_flag" check?)
Fixes: 5f9ae91f7c0d ("kbuild: Build kernel module BTFs if BTF is enabled and pahole supports it")
Signed-off-by: Stanislav Fomichev <sdf@xxxxxxxxxx>
---
lib/Kconfig.debug | 1 +
1 file changed, 1 insertion(+)
diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
index c77fe36bb3d8..6336a697c9f5 100644
--- a/lib/Kconfig.debug
+++ b/lib/Kconfig.debug
@@ -326,6 +326,7 @@ config PAHOLE_HAS_SPLIT_BTF
def_bool $(success, test `$(PAHOLE) --version | sed -E 's/v([0-9]+)\.([0-9]+)/\1\2/'` -ge "119")
config DEBUG_INFO_BTF_MODULES
+ bool "Generate BTF module typeinfo"
def_bool y
depends on DEBUG_INFO_BTF && MODULES && PAHOLE_HAS_SPLIT_BTF
help
--
2.38.0.rc1.362.ged0d419d3c-goog
