* [bpf PATCH] bpf: Do not allow btf_ctx_access with __int128 types
@ 2020-06-24 22:20 John Fastabend
2020-06-25 15:01 ` Daniel Borkmann
0 siblings, 1 reply; 5+ messages in thread
From: John Fastabend @ 2020-06-24 22:20 UTC (permalink / raw)
To: jolsa, andrii.nakryiko, ast, daniel; +Cc: netdev, bpf, john.fastabend
To ensure btf_ctx_access() is safe the verifier checks that the BTF
arg type is an int, enum, or pointer. When the function does the
BTF arg lookup it uses the calculation 'arg = off / 8' using the
fact that registers are 8B. This requires that the first arg is
in the first reg, the second in the second, and so on. However,
for __int128 the arg will consume two registers by default LLVM
implementation. So this will cause the arg layout assumed by the
'arg = off / 8' calculation to be incorrect.
Because __int128 is uncommon this patch applies the easiest fix and
will force int types to be sizeof(u64) or smaller so that they will
fit in a single register.
v2: remove unneeded parens per Andrii's feedback
Fixes: 9e15db66136a1 ("bpf: Implement accurate raw_tp context access via BTF")
Acked-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: John Fastabend <john.fastabend@gmail.com>
---
include/linux/btf.h | 5 +++++
kernel/bpf/btf.c | 4 ++--
2 files changed, 7 insertions(+), 2 deletions(-)
diff --git a/include/linux/btf.h b/include/linux/btf.h
index 5c1ea99..8b81fbb 100644
--- a/include/linux/btf.h
+++ b/include/linux/btf.h
@@ -82,6 +82,11 @@ static inline bool btf_type_is_int(const struct btf_type *t)
return BTF_INFO_KIND(t->info) == BTF_KIND_INT;
}
+static inline bool btf_type_is_small_int(const struct btf_type *t)
+{
+ return btf_type_is_int(t) && t->size <= sizeof(u64);
+}
+
static inline bool btf_type_is_enum(const struct btf_type *t)
{
return BTF_INFO_KIND(t->info) == BTF_KIND_ENUM;
diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c
index 58c9af1..9a1a98d 100644
--- a/kernel/bpf/btf.c
+++ b/kernel/bpf/btf.c
@@ -3746,7 +3746,7 @@ bool btf_ctx_access(int off, int size, enum bpf_access_type type,
return false;
t = btf_type_skip_modifiers(btf, t->type, NULL);
- if (!btf_type_is_int(t)) {
+ if (!btf_type_is_small_int(t)) {
bpf_log(log,
"ret type %s not allowed for fmod_ret\n",
btf_kind_str[BTF_INFO_KIND(t->info)]);
@@ -3768,7 +3768,7 @@ bool btf_ctx_access(int off, int size, enum bpf_access_type type,
/* skip modifiers */
while (btf_type_is_modifier(t))
t = btf_type_by_id(btf, t->type);
- if (btf_type_is_int(t) || btf_type_is_enum(t))
+ if (btf_type_is_small_int(t) || btf_type_is_enum(t))
/* accessing a scalar */
return true;
if (!btf_type_is_ptr(t)) {
^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: [bpf PATCH] bpf: Do not allow btf_ctx_access with __int128 types
2020-06-24 22:20 [bpf PATCH] bpf: Do not allow btf_ctx_access with __int128 types John Fastabend
@ 2020-06-25 15:01 ` Daniel Borkmann
0 siblings, 0 replies; 5+ messages in thread
From: Daniel Borkmann @ 2020-06-25 15:01 UTC (permalink / raw)
To: John Fastabend, jolsa, andrii.nakryiko, ast; +Cc: netdev, bpf
On 6/25/20 12:20 AM, John Fastabend wrote:
> To ensure btf_ctx_access() is safe the verifier checks that the BTF
> arg type is an int, enum, or pointer. When the function does the
> BTF arg lookup it uses the calculation 'arg = off / 8' using the
> fact that registers are 8B. This requires that the first arg is
> in the first reg, the second in the second, and so on. However,
> for __int128 the arg will consume two registers by default LLVM
> implementation. So this will cause the arg layout assumed by the
> 'arg = off / 8' calculation to be incorrect.
>
> Because __int128 is uncommon this patch applies the easiest fix and
> will force int types to be sizeof(u64) or smaller so that they will
> fit in a single register.
>
> v2: remove unneeded parens per Andrii's feedback
>
> Fixes: 9e15db66136a1 ("bpf: Implement accurate raw_tp context access via BTF")
> Acked-by: Andrii Nakryiko <andriin@fb.com>
> Signed-off-by: John Fastabend <john.fastabend@gmail.com>
Applied, thanks!
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [bpf PATCH] bpf: do not allow btf_ctx_access with __int128 types
2020-06-23 18:33 ` Andrii Nakryiko
@ 2020-06-24 22:23 ` John Fastabend
0 siblings, 0 replies; 5+ messages in thread
From: John Fastabend @ 2020-06-24 22:23 UTC (permalink / raw)
To: Andrii Nakryiko, John Fastabend
Cc: Jiri Olsa, Alexei Starovoitov, Daniel Borkmann, Networking, bpf
Andrii Nakryiko wrote:
> On Tue, Jun 23, 2020 at 10:14 AM John Fastabend
> <john.fastabend@gmail.com> wrote:
> >
> > To ensure btf_ctx_access() is safe the verifier checks that the BTF
> > arg type is an int, enum, or pointer. When the function does the
> > BTF arg lookup it uses the calculation 'arg = off / 8' using the
> > fact that registers are 8B. This requires that the first arg is
> > in the first reg, the second in the second, and so on. However,
> > for __int128 the arg will consume two registers by default LLVM
> > implementation. So this will cause the arg layout assumed by the
> > 'arg = off / 8' calculation to be incorrect.
> >
> > Because __int128 is uncommon this patch applies the easiest fix and
> > will force int types to be sizeof(u64) or smaller so that they will
> > fit in a single register.
> >
> > Fixes: 9e15db66136a1 ("bpf: Implement accurate raw_tp context access via BTF")
> > Signed-off-by: John Fastabend <john.fastabend@gmail.com>
> > ---
>
> "small int" for u64 looks funny, but naming is hard :)
>
> Acked-by: Andrii Nakryiko <andriin@fb.com>
>
Fixed up the parens and sent a v2 keeping the ACK. I don't have any
better ideas for the name, let me know if you have a preference for
something else.
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [bpf PATCH] bpf: do not allow btf_ctx_access with __int128 types
2020-06-23 17:13 [bpf PATCH] bpf: do " John Fastabend
@ 2020-06-23 18:33 ` Andrii Nakryiko
2020-06-24 22:23 ` John Fastabend
0 siblings, 1 reply; 5+ messages in thread
From: Andrii Nakryiko @ 2020-06-23 18:33 UTC (permalink / raw)
To: John Fastabend
Cc: Jiri Olsa, Alexei Starovoitov, Daniel Borkmann, Networking, bpf
On Tue, Jun 23, 2020 at 10:14 AM John Fastabend
<john.fastabend@gmail.com> wrote:
>
> To ensure btf_ctx_access() is safe the verifier checks that the BTF
> arg type is an int, enum, or pointer. When the function does the
> BTF arg lookup it uses the calculation 'arg = off / 8' using the
> fact that registers are 8B. This requires that the first arg is
> in the first reg, the second in the second, and so on. However,
> for __int128 the arg will consume two registers by default LLVM
> implementation. So this will cause the arg layout assumed by the
> 'arg = off / 8' calculation to be incorrect.
>
> Because __int128 is uncommon this patch applies the easiest fix and
> will force int types to be sizeof(u64) or smaller so that they will
> fit in a single register.
>
> Fixes: 9e15db66136a1 ("bpf: Implement accurate raw_tp context access via BTF")
> Signed-off-by: John Fastabend <john.fastabend@gmail.com>
> ---
"small int" for u64 looks funny, but naming is hard :)
Acked-by: Andrii Nakryiko <andriin@fb.com>
> include/linux/btf.h | 5 +++++
> kernel/bpf/btf.c | 4 ++--
> 2 files changed, 7 insertions(+), 2 deletions(-)
>
> diff --git a/include/linux/btf.h b/include/linux/btf.h
> index 5c1ea99..35642f6 100644
> --- a/include/linux/btf.h
> +++ b/include/linux/btf.h
> @@ -82,6 +82,11 @@ static inline bool btf_type_is_int(const struct btf_type *t)
> return BTF_INFO_KIND(t->info) == BTF_KIND_INT;
> }
>
> +static inline bool btf_type_is_small_int(const struct btf_type *t)
> +{
> + return btf_type_is_int(t) && (t->size <= sizeof(u64));
nit: unnecessary (), () are usually used to disambiguate | and & vs
|| and &&; this is not the case, though.
> +}
> +
> static inline bool btf_type_is_enum(const struct btf_type *t)
> {
> return BTF_INFO_KIND(t->info) == BTF_KIND_ENUM;
> diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c
> index 58c9af1..9a1a98d 100644
> --- a/kernel/bpf/btf.c
> +++ b/kernel/bpf/btf.c
> @@ -3746,7 +3746,7 @@ bool btf_ctx_access(int off, int size, enum bpf_access_type type,
> return false;
>
> t = btf_type_skip_modifiers(btf, t->type, NULL);
> - if (!btf_type_is_int(t)) {
> + if (!btf_type_is_small_int(t)) {
> bpf_log(log,
> "ret type %s not allowed for fmod_ret\n",
> btf_kind_str[BTF_INFO_KIND(t->info)]);
> @@ -3768,7 +3768,7 @@ bool btf_ctx_access(int off, int size, enum bpf_access_type type,
> /* skip modifiers */
> while (btf_type_is_modifier(t))
> t = btf_type_by_id(btf, t->type);
> - if (btf_type_is_int(t) || btf_type_is_enum(t))
> + if (btf_type_is_small_int(t) || btf_type_is_enum(t))
> /* accessing a scalar */
> return true;
> if (!btf_type_is_ptr(t)) {
>
^ permalink raw reply [flat|nested] 5+ messages in thread
* [bpf PATCH] bpf: do not allow btf_ctx_access with __int128 types
@ 2020-06-23 17:13 John Fastabend
2020-06-23 18:33 ` Andrii Nakryiko
0 siblings, 1 reply; 5+ messages in thread
From: John Fastabend @ 2020-06-23 17:13 UTC (permalink / raw)
To: jolsa, ast, daniel; +Cc: netdev, bpf, john.fastabend
To ensure btf_ctx_access() is safe the verifier checks that the BTF
arg type is an int, enum, or pointer. When the function does the
BTF arg lookup it uses the calculation 'arg = off / 8' using the
fact that registers are 8B. This requires that the first arg is
in the first reg, the second in the second, and so on. However,
for __int128 the arg will consume two registers by default LLVM
implementation. So this will cause the arg layout assumed by the
'arg = off / 8' calculation to be incorrect.
Because __int128 is uncommon this patch applies the easiest fix and
will force int types to be sizeof(u64) or smaller so that they will
fit in a single register.
Fixes: 9e15db66136a1 ("bpf: Implement accurate raw_tp context access via BTF")
Signed-off-by: John Fastabend <john.fastabend@gmail.com>
---
include/linux/btf.h | 5 +++++
kernel/bpf/btf.c | 4 ++--
2 files changed, 7 insertions(+), 2 deletions(-)
diff --git a/include/linux/btf.h b/include/linux/btf.h
index 5c1ea99..35642f6 100644
--- a/include/linux/btf.h
+++ b/include/linux/btf.h
@@ -82,6 +82,11 @@ static inline bool btf_type_is_int(const struct btf_type *t)
return BTF_INFO_KIND(t->info) == BTF_KIND_INT;
}
+static inline bool btf_type_is_small_int(const struct btf_type *t)
+{
+ return btf_type_is_int(t) && (t->size <= sizeof(u64));
+}
+
static inline bool btf_type_is_enum(const struct btf_type *t)
{
return BTF_INFO_KIND(t->info) == BTF_KIND_ENUM;
diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c
index 58c9af1..9a1a98d 100644
--- a/kernel/bpf/btf.c
+++ b/kernel/bpf/btf.c
@@ -3746,7 +3746,7 @@ bool btf_ctx_access(int off, int size, enum bpf_access_type type,
return false;
t = btf_type_skip_modifiers(btf, t->type, NULL);
- if (!btf_type_is_int(t)) {
+ if (!btf_type_is_small_int(t)) {
bpf_log(log,
"ret type %s not allowed for fmod_ret\n",
btf_kind_str[BTF_INFO_KIND(t->info)]);
@@ -3768,7 +3768,7 @@ bool btf_ctx_access(int off, int size, enum bpf_access_type type,
/* skip modifiers */
while (btf_type_is_modifier(t))
t = btf_type_by_id(btf, t->type);
- if (btf_type_is_int(t) || btf_type_is_enum(t))
+ if (btf_type_is_small_int(t) || btf_type_is_enum(t))
/* accessing a scalar */
return true;
if (!btf_type_is_ptr(t)) {
^ permalink raw reply related [flat|nested] 5+ messages in thread
end of thread, other threads:[~2020-06-25 15:01 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-06-24 22:20 [bpf PATCH] bpf: Do not allow btf_ctx_access with __int128 types John Fastabend
2020-06-25 15:01 ` Daniel Borkmann
-- strict thread matches above, loose matches on Subject: below --
2020-06-23 17:13 [bpf PATCH] bpf: do " John Fastabend
2020-06-23 18:33 ` Andrii Nakryiko
2020-06-24 22:23 ` John Fastabend
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.