diff mbox series

[RFC,v2,bpf-next,5/9] selftests/bpf: Add read_buildid function

Message ID 20230228093206.821563-6-jolsa@kernel.org (mailing list archive)
State New, archived
Headers show
Series mm/bpf/perf: Store build id in inode object | expand

Commit Message

Jiri Olsa Feb. 28, 2023, 9:32 a.m. UTC
Adding read_build_id function that parses out build id from
specified binary.

It will replace extract_build_id and also be used in following
changes.

Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
 tools/testing/selftests/bpf/trace_helpers.c | 98 +++++++++++++++++++++
 tools/testing/selftests/bpf/trace_helpers.h |  5 ++
 2 files changed, 103 insertions(+)

Comments

Andrii Nakryiko March 8, 2023, 1:22 a.m. UTC | #1
On Tue, Feb 28, 2023 at 1:33 AM Jiri Olsa <jolsa@kernel.org> wrote:
>
> Adding read_build_id function that parses out build id from
> specified binary.
>
> It will replace extract_build_id and also be used in following
> changes.
>
> Signed-off-by: Jiri Olsa <jolsa@kernel.org>
> ---
>  tools/testing/selftests/bpf/trace_helpers.c | 98 +++++++++++++++++++++
>  tools/testing/selftests/bpf/trace_helpers.h |  5 ++
>  2 files changed, 103 insertions(+)
>
> diff --git a/tools/testing/selftests/bpf/trace_helpers.c b/tools/testing/selftests/bpf/trace_helpers.c
> index 09a16a77bae4..c10e16626cd3 100644
> --- a/tools/testing/selftests/bpf/trace_helpers.c
> +++ b/tools/testing/selftests/bpf/trace_helpers.c
> @@ -11,6 +11,9 @@
>  #include <linux/perf_event.h>
>  #include <sys/mman.h>
>  #include "trace_helpers.h"
> +#include <linux/limits.h>
> +#include <libelf.h>
> +#include <gelf.h>
>
>  #define DEBUGFS "/sys/kernel/debug/tracing/"
>
> @@ -230,3 +233,98 @@ ssize_t get_rel_offset(uintptr_t addr)
>         fclose(f);
>         return -EINVAL;
>  }
> +
> +static int
> +parse_build_id_buf(const void *note_start, Elf32_Word note_size,
> +                  char *build_id)
> +{
> +       Elf32_Word note_offs = 0, new_offs;
> +
> +       while (note_offs + sizeof(Elf32_Nhdr) < note_size) {
> +               Elf32_Nhdr *nhdr = (Elf32_Nhdr *)(note_start + note_offs);
> +
> +               if (nhdr->n_type == 3 &&
> +                   nhdr->n_namesz == sizeof("GNU") &&
> +                   !strcmp((char *)(nhdr + 1), "GNU") &&
> +                   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),
> +                              nhdr->n_descsz);
> +                       memset(build_id + nhdr->n_descsz, 0,
> +                              BPF_BUILD_ID_SIZE - nhdr->n_descsz);

I won't count :) but if something fits within 100 characters, please
keep it on single line

> +                       return (int) nhdr->n_descsz;
> +               }
> +
> +               new_offs = note_offs + sizeof(Elf32_Nhdr) +
> +                          ALIGN(nhdr->n_namesz, 4) + ALIGN(nhdr->n_descsz, 4);
> +
> +               if (new_offs >= note_size)
> +                       break;
> +               note_offs = new_offs;
> +       }
> +
> +       return -EINVAL;
> +}
> +
> +/* Reads binary from *path* file and returns it in the *build_id*
> + * which is expected to be at least BPF_BUILD_ID_SIZE bytes.
> + * Returns size of build id on success. On error the error value
> + * is returned.
> + */
> +int read_build_id(const char *path, char *build_id)
> +{
> +       int fd, err = -EINVAL;
> +       Elf *elf = NULL;
> +       GElf_Ehdr ehdr;
> +       size_t max, i;
> +
> +       fd = open(path, O_RDONLY | O_CLOEXEC);
> +       if (fd < 0)
> +               return -errno;
> +
> +       (void)elf_version(EV_CURRENT);
> +
> +       elf = elf_begin(fd, ELF_C_READ, NULL);
> +       if (!elf)
> +               goto out;
> +
> +       if (elf_kind(elf) != ELF_K_ELF)
> +               goto out;
> +
> +       if (gelf_getehdr(elf, &ehdr) == NULL)
> +               goto out;
> +
> +       if (ehdr.e_ident[EI_CLASS] != ELFCLASS64)
> +               goto out;
> +
> +       for (i = 0; i < ehdr.e_phnum; i++) {
> +               GElf_Phdr mem, *phdr;
> +               char *data;
> +
> +               phdr = gelf_getphdr(elf, i, &mem);
> +               if (!phdr)
> +                       goto out;
> +
> +               if (phdr->p_type != PT_NOTE)
> +                       continue;
> +
> +               data = elf_rawfile(elf, &max);
> +               if (!data)
> +                       goto out;
> +
> +               if (phdr->p_offset >= max ||
> +                  (phdr->p_offset + phdr->p_memsz >= max))
> +                       goto out;
> +
> +               err = parse_build_id_buf(data + phdr->p_offset, phdr->p_memsz, build_id);
> +               if (err > 0)
> +                       goto out;
> +       }
> +
> +out:
> +       if (elf)
> +               elf_end(elf);
> +       close(fd);
> +       return err;
> +}
> diff --git a/tools/testing/selftests/bpf/trace_helpers.h b/tools/testing/selftests/bpf/trace_helpers.h
> index 53efde0e2998..50b2cc498ba7 100644
> --- a/tools/testing/selftests/bpf/trace_helpers.h
> +++ b/tools/testing/selftests/bpf/trace_helpers.h
> @@ -4,6 +4,9 @@
>
>  #include <bpf/libbpf.h>
>
> +#define ALIGN(x, a)            __ALIGN_MASK(x, (typeof(x))(a)-1)
> +#define __ALIGN_MASK(x, mask)  (((x)+(mask))&~(mask))

nit: I know these are macros, but why would you first use __ALIGN_MASK
and then #define it? swap them?


> +
>  struct ksym {
>         long addr;
>         char *name;
> @@ -23,4 +26,6 @@ void read_trace_pipe(void);
>  ssize_t get_uprobe_offset(const void *addr);
>  ssize_t get_rel_offset(uintptr_t addr);
>
> +int read_build_id(const char *path, char *build_id);
> +
>  #endif
> --
> 2.39.2
>
Jiri Olsa March 8, 2023, 1:49 p.m. UTC | #2
On Tue, Mar 07, 2023 at 05:22:51PM -0800, Andrii Nakryiko wrote:
> On Tue, Feb 28, 2023 at 1:33 AM Jiri Olsa <jolsa@kernel.org> wrote:
> >
> > Adding read_build_id function that parses out build id from
> > specified binary.
> >
> > It will replace extract_build_id and also be used in following
> > changes.
> >
> > Signed-off-by: Jiri Olsa <jolsa@kernel.org>
> > ---
> >  tools/testing/selftests/bpf/trace_helpers.c | 98 +++++++++++++++++++++
> >  tools/testing/selftests/bpf/trace_helpers.h |  5 ++
> >  2 files changed, 103 insertions(+)
> >
> > diff --git a/tools/testing/selftests/bpf/trace_helpers.c b/tools/testing/selftests/bpf/trace_helpers.c
> > index 09a16a77bae4..c10e16626cd3 100644
> > --- a/tools/testing/selftests/bpf/trace_helpers.c
> > +++ b/tools/testing/selftests/bpf/trace_helpers.c
> > @@ -11,6 +11,9 @@
> >  #include <linux/perf_event.h>
> >  #include <sys/mman.h>
> >  #include "trace_helpers.h"
> > +#include <linux/limits.h>
> > +#include <libelf.h>
> > +#include <gelf.h>
> >
> >  #define DEBUGFS "/sys/kernel/debug/tracing/"
> >
> > @@ -230,3 +233,98 @@ ssize_t get_rel_offset(uintptr_t addr)
> >         fclose(f);
> >         return -EINVAL;
> >  }
> > +
> > +static int
> > +parse_build_id_buf(const void *note_start, Elf32_Word note_size,
> > +                  char *build_id)
> > +{
> > +       Elf32_Word note_offs = 0, new_offs;
> > +
> > +       while (note_offs + sizeof(Elf32_Nhdr) < note_size) {
> > +               Elf32_Nhdr *nhdr = (Elf32_Nhdr *)(note_start + note_offs);
> > +
> > +               if (nhdr->n_type == 3 &&
> > +                   nhdr->n_namesz == sizeof("GNU") &&
> > +                   !strcmp((char *)(nhdr + 1), "GNU") &&
> > +                   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),
> > +                              nhdr->n_descsz);
> > +                       memset(build_id + nhdr->n_descsz, 0,
> > +                              BPF_BUILD_ID_SIZE - nhdr->n_descsz);
> 
> I won't count :) but if something fits within 100 characters, please
> keep it on single line

copy&paste from kernel code ;-) I'll reformat that

SNIP

> > diff --git a/tools/testing/selftests/bpf/trace_helpers.h b/tools/testing/selftests/bpf/trace_helpers.h
> > index 53efde0e2998..50b2cc498ba7 100644
> > --- a/tools/testing/selftests/bpf/trace_helpers.h
> > +++ b/tools/testing/selftests/bpf/trace_helpers.h
> > @@ -4,6 +4,9 @@
> >
> >  #include <bpf/libbpf.h>
> >
> > +#define ALIGN(x, a)            __ALIGN_MASK(x, (typeof(x))(a)-1)
> > +#define __ALIGN_MASK(x, mask)  (((x)+(mask))&~(mask))
> 
> nit: I know these are macros, but why would you first use __ALIGN_MASK
> and then #define it? swap them?

same reason as above, I'll swap that

thanks,
jirka
diff mbox series

Patch

diff --git a/tools/testing/selftests/bpf/trace_helpers.c b/tools/testing/selftests/bpf/trace_helpers.c
index 09a16a77bae4..c10e16626cd3 100644
--- a/tools/testing/selftests/bpf/trace_helpers.c
+++ b/tools/testing/selftests/bpf/trace_helpers.c
@@ -11,6 +11,9 @@ 
 #include <linux/perf_event.h>
 #include <sys/mman.h>
 #include "trace_helpers.h"
+#include <linux/limits.h>
+#include <libelf.h>
+#include <gelf.h>
 
 #define DEBUGFS "/sys/kernel/debug/tracing/"
 
@@ -230,3 +233,98 @@  ssize_t get_rel_offset(uintptr_t addr)
 	fclose(f);
 	return -EINVAL;
 }
+
+static int
+parse_build_id_buf(const void *note_start, Elf32_Word note_size,
+		   char *build_id)
+{
+	Elf32_Word note_offs = 0, new_offs;
+
+	while (note_offs + sizeof(Elf32_Nhdr) < note_size) {
+		Elf32_Nhdr *nhdr = (Elf32_Nhdr *)(note_start + note_offs);
+
+		if (nhdr->n_type == 3 &&
+		    nhdr->n_namesz == sizeof("GNU") &&
+		    !strcmp((char *)(nhdr + 1), "GNU") &&
+		    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),
+			       nhdr->n_descsz);
+			memset(build_id + nhdr->n_descsz, 0,
+			       BPF_BUILD_ID_SIZE - nhdr->n_descsz);
+			return (int) nhdr->n_descsz;
+		}
+
+		new_offs = note_offs + sizeof(Elf32_Nhdr) +
+			   ALIGN(nhdr->n_namesz, 4) + ALIGN(nhdr->n_descsz, 4);
+
+		if (new_offs >= note_size)
+			break;
+		note_offs = new_offs;
+	}
+
+	return -EINVAL;
+}
+
+/* Reads binary from *path* file and returns it in the *build_id*
+ * which is expected to be at least BPF_BUILD_ID_SIZE bytes.
+ * Returns size of build id on success. On error the error value
+ * is returned.
+ */
+int read_build_id(const char *path, char *build_id)
+{
+	int fd, err = -EINVAL;
+	Elf *elf = NULL;
+	GElf_Ehdr ehdr;
+	size_t max, i;
+
+	fd = open(path, O_RDONLY | O_CLOEXEC);
+	if (fd < 0)
+		return -errno;
+
+	(void)elf_version(EV_CURRENT);
+
+	elf = elf_begin(fd, ELF_C_READ, NULL);
+	if (!elf)
+		goto out;
+
+	if (elf_kind(elf) != ELF_K_ELF)
+		goto out;
+
+	if (gelf_getehdr(elf, &ehdr) == NULL)
+		goto out;
+
+	if (ehdr.e_ident[EI_CLASS] != ELFCLASS64)
+		goto out;
+
+	for (i = 0; i < ehdr.e_phnum; i++) {
+		GElf_Phdr mem, *phdr;
+		char *data;
+
+		phdr = gelf_getphdr(elf, i, &mem);
+		if (!phdr)
+			goto out;
+
+		if (phdr->p_type != PT_NOTE)
+			continue;
+
+		data = elf_rawfile(elf, &max);
+		if (!data)
+			goto out;
+
+		if (phdr->p_offset >= max ||
+		   (phdr->p_offset + phdr->p_memsz >= max))
+			goto out;
+
+		err = parse_build_id_buf(data + phdr->p_offset, phdr->p_memsz, build_id);
+		if (err > 0)
+			goto out;
+	}
+
+out:
+	if (elf)
+		elf_end(elf);
+	close(fd);
+	return err;
+}
diff --git a/tools/testing/selftests/bpf/trace_helpers.h b/tools/testing/selftests/bpf/trace_helpers.h
index 53efde0e2998..50b2cc498ba7 100644
--- a/tools/testing/selftests/bpf/trace_helpers.h
+++ b/tools/testing/selftests/bpf/trace_helpers.h
@@ -4,6 +4,9 @@ 
 
 #include <bpf/libbpf.h>
 
+#define ALIGN(x, a)		__ALIGN_MASK(x, (typeof(x))(a)-1)
+#define __ALIGN_MASK(x, mask)	(((x)+(mask))&~(mask))
+
 struct ksym {
 	long addr;
 	char *name;
@@ -23,4 +26,6 @@  void read_trace_pipe(void);
 ssize_t get_uprobe_offset(const void *addr);
 ssize_t get_rel_offset(uintptr_t addr);
 
+int read_build_id(const char *path, char *build_id);
+
 #endif