diff mbox series

[bpf,v2,1/3] bpf: don't assume build-id length is always 20 bytes

Message ID 20190116181119.96117-1-sdf@google.com
State Changes Requested
Delegated to: BPF Maintainers
Headers show
Series [bpf,v2,1/3] bpf: don't assume build-id length is always 20 bytes | expand

Commit Message

Stanislav Fomichev Jan. 16, 2019, 6:11 p.m. UTC
Build-id length is not fixed to 20, it can be (`man ld` /--build-id):
  * 128-bit (uuid)
  * 160-bit (sha1)
  * any length specified in ld --build-id=0xhexstring

To fix the issue of missing BPF_STACK_BUILD_ID_VALID for shorter build-ids,
assume that build-id is somewhere in the range of 1 .. 20.
Set the remaining bytes to zero.

v2:
* don't introduce new "len = min(BPF_BUILD_ID_SIZE, nhdr->n_descsz)",
  we already know that nhdr->n_descsz <= BPF_BUILD_ID_SIZE if we enter
  this 'if' condition

Fixes: 615755a77b24 ("bpf: extend stackmap to save binary_build_id+offset instead of address")
Signed-off-by: Stanislav Fomichev <sdf@google.com>
---
 kernel/bpf/stackmap.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

Comments

Song Liu Jan. 16, 2019, 6:20 p.m. UTC | #1
> On Jan 16, 2019, at 10:11 AM, Stanislav Fomichev <sdf@google.com> wrote:
> 
> Build-id length is not fixed to 20, it can be (`man ld` /--build-id):
>  * 128-bit (uuid)
>  * 160-bit (sha1)
>  * any length specified in ld --build-id=0xhexstring
> 
> To fix the issue of missing BPF_STACK_BUILD_ID_VALID for shorter build-ids,
> assume that build-id is somewhere in the range of 1 .. 20.
> Set the remaining bytes to zero.
> 
> v2:
> * don't introduce new "len = min(BPF_BUILD_ID_SIZE, nhdr->n_descsz)",
>  we already know that nhdr->n_descsz <= BPF_BUILD_ID_SIZE if we enter
>  this 'if' condition
> 
> Fixes: 615755a77b24 ("bpf: extend stackmap to save binary_build_id+offset instead of address")
> Signed-off-by: Stanislav Fomichev <sdf@google.com>

Acked-by: Song Liu <songliubraving@fb.com>
 

> ---
> kernel/bpf/stackmap.c | 7 +++++--
> 1 file changed, 5 insertions(+), 2 deletions(-)
> 
> diff --git a/kernel/bpf/stackmap.c b/kernel/bpf/stackmap.c
> index d9e2483669d0..f9df545e92f6 100644
> --- a/kernel/bpf/stackmap.c
> +++ b/kernel/bpf/stackmap.c
> @@ -180,11 +180,14 @@ static inline int stack_map_parse_build_id(void *page_addr,
> 
> 		if (nhdr->n_type == BPF_BUILD_ID &&
> 		    nhdr->n_namesz == sizeof("GNU") &&
> -		    nhdr->n_descsz == BPF_BUILD_ID_SIZE) {
> +		    nhdr->n_descsz > 0 &&
> +		    nhdr->n_descsz <= BPF_BUILD_ID_SIZE) {
> 			memcpy(build_id,
> 			       note_start + note_offs +
> 			       ALIGN(sizeof("GNU"), 4) + sizeof(Elf32_Nhdr),
> -			       BPF_BUILD_ID_SIZE);
> +			       nhdr->n_descsz);
> +			memset(build_id + nhdr->n_descsz, 0,
> +			       BPF_BUILD_ID_SIZE - nhdr->n_descsz);
> 			return 0;
> 		}
> 		new_offs = note_offs + sizeof(Elf32_Nhdr) +
> -- 
> 2.20.1.97.g81188d93c3-goog
>
Daniel Borkmann Jan. 16, 2019, 9:59 p.m. UTC | #2
On 01/16/2019 07:11 PM, Stanislav Fomichev wrote:
> Build-id length is not fixed to 20, it can be (`man ld` /--build-id):
>   * 128-bit (uuid)
>   * 160-bit (sha1)
>   * any length specified in ld --build-id=0xhexstring
> 
> To fix the issue of missing BPF_STACK_BUILD_ID_VALID for shorter build-ids,
> assume that build-id is somewhere in the range of 1 .. 20.
> Set the remaining bytes to zero.
> 
> v2:
> * don't introduce new "len = min(BPF_BUILD_ID_SIZE, nhdr->n_descsz)",
>   we already know that nhdr->n_descsz <= BPF_BUILD_ID_SIZE if we enter
>   this 'if' condition
> 
> Fixes: 615755a77b24 ("bpf: extend stackmap to save binary_build_id+offset instead of address")
> Signed-off-by: Stanislav Fomichev <sdf@google.com>

Hmm, looks like rest of the v2 series didn't make it to the list.
Please double check; just in case simply resend the full v2 set so
it properly ends up in patchwork.

Thanks,
Daniel
Stanislav Fomichev Jan. 16, 2019, 10:01 p.m. UTC | #3
On 01/16, Daniel Borkmann wrote:
> On 01/16/2019 07:11 PM, Stanislav Fomichev wrote:
> > Build-id length is not fixed to 20, it can be (`man ld` /--build-id):
> >   * 128-bit (uuid)
> >   * 160-bit (sha1)
> >   * any length specified in ld --build-id=0xhexstring
> > 
> > To fix the issue of missing BPF_STACK_BUILD_ID_VALID for shorter build-ids,
> > assume that build-id is somewhere in the range of 1 .. 20.
> > Set the remaining bytes to zero.
> > 
> > v2:
> > * don't introduce new "len = min(BPF_BUILD_ID_SIZE, nhdr->n_descsz)",
> >   we already know that nhdr->n_descsz <= BPF_BUILD_ID_SIZE if we enter
> >   this 'if' condition
> > 
> > Fixes: 615755a77b24 ("bpf: extend stackmap to save binary_build_id+offset instead of address")
> > Signed-off-by: Stanislav Fomichev <sdf@google.com>
> 
> Hmm, looks like rest of the v2 series didn't make it to the list.
> Please double check; just in case simply resend the full v2 set so
> it properly ends up in patchwork.
Oh, I didn't send the first two patches because I didn't change them, I'll
resend full v2 series in a moment.
> 
> Thanks,
> Daniel
diff mbox series

Patch

diff --git a/kernel/bpf/stackmap.c b/kernel/bpf/stackmap.c
index d9e2483669d0..f9df545e92f6 100644
--- a/kernel/bpf/stackmap.c
+++ b/kernel/bpf/stackmap.c
@@ -180,11 +180,14 @@  static inline int stack_map_parse_build_id(void *page_addr,
 
 		if (nhdr->n_type == BPF_BUILD_ID &&
 		    nhdr->n_namesz == sizeof("GNU") &&
-		    nhdr->n_descsz == BPF_BUILD_ID_SIZE) {
+		    nhdr->n_descsz > 0 &&
+		    nhdr->n_descsz <= BPF_BUILD_ID_SIZE) {
 			memcpy(build_id,
 			       note_start + note_offs +
 			       ALIGN(sizeof("GNU"), 4) + sizeof(Elf32_Nhdr),
-			       BPF_BUILD_ID_SIZE);
+			       nhdr->n_descsz);
+			memset(build_id + nhdr->n_descsz, 0,
+			       BPF_BUILD_ID_SIZE - nhdr->n_descsz);
 			return 0;
 		}
 		new_offs = note_offs + sizeof(Elf32_Nhdr) +