Skip to contentSkip to conversation

libbpf: skip base btf sanity checks

PATCH Applied Antoine Tenart 4 participants 4 messages Jun 24, 2024

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 libbpf: skip base btf sanity checks

When upgrading to libbpf 1.3 we noticed a big performance hit while loading programs using CORE on non base-BTF symbols. This was tracked down to the new BTF sanity check logic. The issue is the base BTF definitions are checked first for the base BTF and then again for every module BTF.

Loading 5 dummy programs (using libbpf-rs) that are using CORE on a non-base BTF symbol on my system:
- Before this fix: 3s.
- With this fix: 0.1s.

Fix this by only checking the types starting at the BTF start id. This should ensure the base BTF is still checked as expected but only once (btf->start_id == 1 when creating the base BTF), and then only additional types are checked for each module BTF.

Fixes 3903802bb99a ("libbpf: Add basic BTF sanity validation")
---
 tools/lib/bpf/btf.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/lib/bpf/btf.c +1 -1
index 2d0840ef599a..142060bbce0a 100644
@@ -598,7 +598,7 @@ static int btf_sanity_check(const struct btf *btf)
 	__u32 i, n = btf__type_cnt(btf);
 	int err;
 
-	for (i = 1; i < n; i++) {
+	for (i = btf->start_id; i < n; i++) {
 		t = btf_type_by_id(btf, i);
 		err = btf_validate_type(btf, t, i);
 		if (err)
Signature · 1 line
2.45.2
Reviewed-by on v1 1/1
#1 lore raw
Quoted 11 lines · Antoine Tenart

When upgrading to libbpf 1.3 we noticed a big performance hit while loading programs using CORE on non base-BTF symbols. This was tracked down to the new BTF sanity check logic. The issue is the base BTF definitions are checked first for the base BTF and then again for every module BTF.

Loading 5 dummy programs (using libbpf-rs) that are using CORE on a non-base BTF symbol on my system:
- Before this fix: 3s.
- With this fix: 0.1s.

Quoted 7 lines

Fix this by only checking the types starting at the BTF start id. This should ensure the base BTF is still checked as expected but only once (btf->start_id == 1 when creating the base BTF), and then only additional types are checked for each module BTF.

Fixes 3903802bb99a ("libbpf: Add basic BTF sanity validation")

This looks good to me.

Quoted 17 lines
---
 tools/lib/bpf/btf.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/lib/bpf/btf.c +1 -1
index 2d0840ef599a..142060bbce0a 100644
@@ -598,7 +598,7 @@ static int btf_sanity_check(const struct btf *btf)
 	__u32 i, n = btf__type_cnt(btf);
 	int err;
 
-	for (i = 1; i < n; i++) {
+	for (i = btf->start_id; i < n; i++) {
 		t = btf_type_by_id(btf, i);
 		err = btf_validate_type(btf, t, i);
 		if (err)
#2 lore raw
In reply to Antoine Tenart
Quoted 17 lines · Antoine Tenart

When upgrading to libbpf 1.3 we noticed a big performance hit while loading programs using CORE on non base-BTF symbols. This was tracked down to the new BTF sanity check logic. The issue is the base BTF definitions are checked first for the base BTF and then again for every module BTF.

Loading 5 dummy programs (using libbpf-rs) that are using CORE on a non-base BTF symbol on my system:
- Before this fix: 3s.
- With this fix: 0.1s.

Fix this by only checking the types starting at the BTF start id. This should ensure the base BTF is still checked as expected but only once (btf->start_id == 1 when creating the base BTF), and then only additional types are checked for each module BTF.

Quoted 6 lines
Fixes 3903802bb99a ("libbpf: Add basic BTF sanity validation")
---
 tools/lib/bpf/btf.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Makes total sense, thanks, applied!

Quoted 16 lines
tools/lib/bpf/btf.c +1 -2
index 2d0840ef599a..142060bbce0a 100644
@@ -598,7 +598,7 @@ static int btf_sanity_check(const struct btf *btf)
        __u32 i, n = btf__type_cnt(btf);
        int err;
 
-       for (i = 1; i < n; i++) {
+       for (i = btf->start_id; i < n; i++) {
                t = btf_type_by_id(btf, i);
                err = btf_validate_type(btf, t, i);
                if (err)
--

2.45.2

#3 lore raw
In reply to Antoine Tenart
This patch was applied to bpf/bpf-next.git (master) by Andrii Nakryiko <[email protected]>:

Hello:

This patch was applied to bpf/bpf-next.git (master) by Andrii Nakryiko <[email protected]>:

Quoted 6 lines · On Mon, 24 Jun 2024 11:09:07 +0200 you wrote:

When upgrading to libbpf 1.3 we noticed a big performance hit while loading programs using CORE on non base-BTF symbols. This was tracked down to the new BTF sanity check logic. The issue is the base BTF definitions are checked first for the base BTF and then again for every module BTF.

Quoted 6 lines

Loading 5 dummy programs (using libbpf-rs) that are using CORE on a non-base BTF symbol on my system:
- Before this fix: 3s.
- With this fix: 0.1s.

[...]

Here is the summary with links:
- [bpf] libbpf: skip base btf sanity checks

    https://git.kernel.org/bpf/bpf-next/c/c73a9683cb21

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html