linux.git/tools/lib/bpf/btf_dump.c, branch v5.9 Linux kernel source tree libbpf: Fix build on ppc64le architecture 2020-08-18T17:02:52+00:00 Andrii Nakryiko andriin@fb.com 2020-08-18T16:44:56+00:00 3fb1a96a91120877488071a167d26d76be4be977 On ppc64le we get the following warning: In file included from btf_dump.c:16:0: btf_dump.c: In function ‘btf_dump_emit_struct_def’: ../include/linux/kernel.h:20:17: error: comparison of distinct pointer types lacks a cast [-Werror] (void) (&_max1 == &_max2); \ ^ btf_dump.c:882:11: note: in expansion of macro ‘max’ m_sz = max(0LL, btf__resolve_size(d->btf, m->type)); ^~~ Fix by explicitly casting to __s64, which is a return type from btf__resolve_size(). Fixes: 702eddc77a90 ("libbpf: Handle GCC built-in types for Arm NEON") Signed-off-by: Andrii Nakryiko <andriin@fb.com> Signed-off-by: Alexei Starovoitov <ast@kernel.org> Link: https://lore.kernel.org/bpf/20200818164456.1181661-1-andriin@fb.com 
On ppc64le we get the following warning:

  In file included from btf_dump.c:16:0:
  btf_dump.c: In function ‘btf_dump_emit_struct_def’:
  ../include/linux/kernel.h:20:17: error: comparison of distinct pointer types lacks a cast [-Werror]
    (void) (&_max1 == &_max2);  \
                   ^
  btf_dump.c:882:11: note: in expansion of macro ‘max’
      m_sz = max(0LL, btf__resolve_size(d->btf, m->type));
             ^~~

Fix by explicitly casting to __s64, which is a return type from
btf__resolve_size().

Fixes: 702eddc77a90 ("libbpf: Handle GCC built-in types for Arm NEON")
Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200818164456.1181661-1-andriin@fb.com
libbpf: Handle BTF pointer sizes more carefully 2020-08-13T23:45:41+00:00 Andrii Nakryiko andriin@fb.com 2020-08-13T20:49:40+00:00 44ad23dfbccbcd26d6ca504eba1ac55755864969 With libbpf and BTF it is pretty common to have libbpf built for one architecture, while BTF information was generated for a different architecture (typically, but not always, BPF). In such case, the size of a pointer might differ betweem architectures. libbpf previously was always making an assumption that pointer size for BTF is the same as native architecture pointer size, but that breaks for cases where libbpf is built as 32-bit library, while BTF is for 64-bit architecture. To solve this, add heuristic to determine pointer size by searching for `long` or `unsigned long` integer type and using its size as a pointer size. Also, allow to override the pointer size with a new API btf__set_pointer_size(), for cases where application knows which pointer size should be used. User application can check what libbpf "guessed" by looking at the result of btf__pointer_size(). If it's not 0, then libbpf successfully determined a pointer size, otherwise native arch pointer size will be used. For cases where BTF is parsed from ELF file, use ELF's class (32-bit or 64-bit) to determine pointer size. Fixes: 8a138aed4a80 ("bpf: btf: Add BTF support to libbpf") Fixes: 351131b51c7a ("libbpf: add btf_dump API for BTF-to-C conversion") Signed-off-by: Andrii Nakryiko <andriin@fb.com> Signed-off-by: Alexei Starovoitov <ast@kernel.org> Link: https://lore.kernel.org/bpf/20200813204945.1020225-5-andriin@fb.com 
With libbpf and BTF it is pretty common to have libbpf built for one
architecture, while BTF information was generated for a different architecture
(typically, but not always, BPF). In such case, the size of a pointer might
differ betweem architectures. libbpf previously was always making an
assumption that pointer size for BTF is the same as native architecture
pointer size, but that breaks for cases where libbpf is built as 32-bit
library, while BTF is for 64-bit architecture.

To solve this, add heuristic to determine pointer size by searching for `long`
or `unsigned long` integer type and using its size as a pointer size. Also,
allow to override the pointer size with a new API btf__set_pointer_size(), for
cases where application knows which pointer size should be used. User
application can check what libbpf "guessed" by looking at the result of
btf__pointer_size(). If it's not 0, then libbpf successfully determined a
pointer size, otherwise native arch pointer size will be used.

For cases where BTF is parsed from ELF file, use ELF's class (32-bit or
64-bit) to determine pointer size.

Fixes: 8a138aed4a80 ("bpf: btf: Add BTF support to libbpf")
Fixes: 351131b51c7a ("libbpf: add btf_dump API for BTF-to-C conversion")
Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200813204945.1020225-5-andriin@fb.com
libbpf: Handle GCC built-in types for Arm NEON 2020-08-13T01:11:51+00:00 Jean-Philippe Brucker jean-philippe@linaro.org 2020-08-12T14:39:10+00:00 702eddc77a905782083b14ccd05b23840675fd18 When building Arm NEON (SIMD) code from lib/raid6/neon.uc, GCC emits DWARF information using a base type "__Poly8_t", which is internal to GCC and not recognized by Clang. This causes build failures when building with Clang a vmlinux.h generated from an arm64 kernel that was built with GCC. vmlinux.h:47284:9: error: unknown type name '__Poly8_t' typedef __Poly8_t poly8x16_t[16]; ^~~~~~~~~ The polyX_t types are defined as unsigned integers in the "Arm C Language Extension" document (101028_Q220_00_en). Emit typedefs based on standard integer types for the GCC internal types, similar to those emitted by Clang. Including linux/kernel.h to use ARRAY_SIZE() incidentally redefined max(), causing a build bug due to different types, hence the seemingly unrelated change. Reported-by: Jakov Petrina <jakov.petrina@sartura.hr> Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org> Signed-off-by: Alexei Starovoitov <ast@kernel.org> Acked-by: Andrii Nakryiko <andriin@fb.com> Link: https://lore.kernel.org/bpf/20200812143909.3293280-1-jean-philippe@linaro.org 
When building Arm NEON (SIMD) code from lib/raid6/neon.uc, GCC emits
DWARF information using a base type "__Poly8_t", which is internal to
GCC and not recognized by Clang. This causes build failures when
building with Clang a vmlinux.h generated from an arm64 kernel that was
built with GCC.

	vmlinux.h:47284:9: error: unknown type name '__Poly8_t'
	typedef __Poly8_t poly8x16_t[16];
	        ^~~~~~~~~

The polyX_t types are defined as unsigned integers in the "Arm C
Language Extension" document (101028_Q220_00_en). Emit typedefs based on
standard integer types for the GCC internal types, similar to those
emitted by Clang.

Including linux/kernel.h to use ARRAY_SIZE() incidentally redefined
max(), causing a build bug due to different types, hence the seemingly
unrelated change.

Reported-by: Jakov Petrina <jakov.petrina@sartura.hr>
Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Andrii Nakryiko <andriin@fb.com>
Link: https://lore.kernel.org/bpf/20200812143909.3293280-1-jean-philippe@linaro.org
libbpf: Make destructors more robust by handling ERR_PTR(err) cases 2020-07-30T22:53:07+00:00 Andrii Nakryiko andriin@fb.com 2020-07-29T23:21:48+00:00 50450fc716c1a570ee8d8bfe198ef5d3cfca36e4 Most of libbpf "constructors" on failure return ERR_PTR(err) result encoded as a pointer. It's a common mistake to eventually pass such malformed pointers into xxx__destroy()/xxx__free() "destructors". So instead of fixing up clean up code in selftests and user programs, handle such error pointers in destructors themselves. This works beautifully for NULL pointers passed to destructors, so might as well just work for error pointers. Suggested-by: Song Liu <songliubraving@fb.com> Signed-off-by: Andrii Nakryiko <andriin@fb.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Acked-by: Song Liu <songliubraving@fb.com> Link: https://lore.kernel.org/bpf/20200729232148.896125-1-andriin@fb.com 
Most of libbpf "constructors" on failure return ERR_PTR(err) result encoded as
a pointer. It's a common mistake to eventually pass such malformed pointers
into xxx__destroy()/xxx__free() "destructors". So instead of fixing up
clean up code in selftests and user programs, handle such error pointers in
destructors themselves. This works beautifully for NULL pointers passed to
destructors, so might as well just work for error pointers.

Suggested-by: Song Liu <songliubraving@fb.com>
Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Song Liu <songliubraving@fb.com>
Link: https://lore.kernel.org/bpf/20200729232148.896125-1-andriin@fb.com
libbpf: Support stripping modifiers for btf_dump 2020-07-14T00:07:43+00:00 Andrii Nakryiko andriin@fb.com 2020-07-13T23:24:08+00:00 7c819e701382d4969ca4837b8cbe157895f5d0bf One important use case when emitting const/volatile/restrict is undesirable is BPF skeleton generation of DATASEC layout. These are further memory-mapped and can be written/read from user-space directly. For important case of .rodata variables, bpftool strips away first-level modifiers, to make their use on user-space side simple and not requiring extra type casts to override compiler complaining about writing to const variables. This logic works mostly fine, but breaks in some more complicated cases. E.g.: const volatile int params[10]; Because in BTF it's a chain of ARRAY -> CONST -> VOLATILE -> INT, bpftool stops at ARRAY and doesn't strip CONST and VOLATILE. In skeleton this variable will be emitted as is. So when used from user-space, compiler will complain about writing to const array. This is problematic, as also mentioned in [0]. To solve this for arrays and other non-trivial cases (e.g., inner const/volatile fields inside the struct), teach btf_dump to strip away any modifier, when requested. This is done as an extra option on btf_dump__emit_type_decl() API. Reported-by: Anton Protopopov <a.s.protopopov@gmail.com> Signed-off-by: Andrii Nakryiko <andriin@fb.com> Signed-off-by: Alexei Starovoitov <ast@kernel.org> Link: https://lore.kernel.org/bpf/20200713232409.3062144-2-andriin@fb.com 
One important use case when emitting const/volatile/restrict is undesirable is
BPF skeleton generation of DATASEC layout. These are further memory-mapped and
can be written/read from user-space directly.

For important case of .rodata variables, bpftool strips away first-level
modifiers, to make their use on user-space side simple and not requiring extra
type casts to override compiler complaining about writing to const variables.

This logic works mostly fine, but breaks in some more complicated cases. E.g.:

    const volatile int params[10];

Because in BTF it's a chain of ARRAY -> CONST -> VOLATILE -> INT, bpftool
stops at ARRAY and doesn't strip CONST and VOLATILE. In skeleton this variable
will be emitted as is. So when used from user-space, compiler will complain
about writing to const array. This is problematic, as also mentioned in [0].

To solve this for arrays and other non-trivial cases (e.g., inner
const/volatile fields inside the struct), teach btf_dump to strip away any
modifier, when requested. This is done as an extra option on
btf_dump__emit_type_decl() API.

Reported-by: Anton Protopopov <a.s.protopopov@gmail.com>
Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200713232409.3062144-2-andriin@fb.com
libbpf: Handle GCC noreturn-turned-volatile quirk 2020-06-10T11:37:02+00:00 Andrii Nakryiko andriin@fb.com 2020-06-10T05:23:35+00:00 32022fd97ed34f6812802bf1288db27c313576f4 Handle a GCC quirk of emitting extra volatile modifier in DWARF (and subsequently preserved in BTF by pahole) for function pointers marked as __attribute__((noreturn)). This was the way to mark such functions before GCC 2.5 added noreturn attribute. Drop such func_proto modifiers, similarly to how it's done for array (also to handle GCC quirk/bug). Such volatile attribute is emitted by GCC only, so existing selftests can't express such test. Simple repro is like this (compiled with GCC + BTF generated by pahole): struct my_struct { void __attribute__((noreturn)) (*fn)(int); }; struct my_struct a; Without this fix, output will be: struct my_struct { voidvolatile (*fn)(int); }; With the fix: struct my_struct { void (*fn)(int); }; Fixes: 351131b51c7a ("libbpf: add btf_dump API for BTF-to-C conversion") Reported-by: Jean-Philippe Brucker <jean-philippe@linaro.org> Signed-off-by: Andrii Nakryiko <andriin@fb.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Tested-by: Jean-Philippe Brucker <jean-philippe@linaro.org> Link: https://lore.kernel.org/bpf/20200610052335.2862559-1-andriin@fb.com 
Handle a GCC quirk of emitting extra volatile modifier in DWARF (and
subsequently preserved in BTF by pahole) for function pointers marked as
__attribute__((noreturn)). This was the way to mark such functions before GCC
2.5 added noreturn attribute. Drop such func_proto modifiers, similarly to how
it's done for array (also to handle GCC quirk/bug).

Such volatile attribute is emitted by GCC only, so existing selftests can't
express such test. Simple repro is like this (compiled with GCC + BTF
generated by pahole):

  struct my_struct {
      void __attribute__((noreturn)) (*fn)(int);
  };
  struct my_struct a;

Without this fix, output will be:

struct my_struct {
    voidvolatile  (*fn)(int);
};

With the fix:

struct my_struct {
    void (*fn)(int);
};

Fixes: 351131b51c7a ("libbpf: add btf_dump API for BTF-to-C conversion")
Reported-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Tested-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Link: https://lore.kernel.org/bpf/20200610052335.2862559-1-andriin@fb.com
libbpf: Remove unneeded semicolon in btf_dump_emit_type 2020-04-28T19:47:47+00:00 Zou Wei zou_wei@huawei.com 2020-04-28T09:07:09+00:00 a6bbdf2e750f245d219d39f3c3d06ace2c5871e6 Fixes the following coccicheck warning: tools/lib/bpf/btf_dump.c:661:4-5: Unneeded semicolon Reported-by: Hulk Robot <hulkci@huawei.com> Signed-off-by: Zou Wei <zou_wei@huawei.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Link: https://lore.kernel.org/bpf/1588064829-70613-1-git-send-email-zou_wei@huawei.com 
Fixes the following coccicheck warning:

 tools/lib/bpf/btf_dump.c:661:4-5: Unneeded semicolon

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Zou Wei <zou_wei@huawei.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/1588064829-70613-1-git-send-email-zou_wei@huawei.com
libbpf: Assume unsigned values for BTF_KIND_ENUM 2020-03-04T16:00:06+00:00 Andrii Nakryiko andriin@fb.com 2020-03-03T00:32:32+00:00 7cb30aaab3f277aa88e20a008faf57e0fb1119ec Currently, BTF_KIND_ENUM type doesn't record whether enum values should be interpreted as signed or unsigned. In Linux, most enums are unsigned, though, so interpreting them as unsigned matches real world better. Change btf_dump test case to test maximum 32-bit value, instead of negative value. Signed-off-by: Andrii Nakryiko <andriin@fb.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Link: https://lore.kernel.org/bpf/20200303003233.3496043-3-andriin@fb.com 
Currently, BTF_KIND_ENUM type doesn't record whether enum values should be
interpreted as signed or unsigned. In Linux, most enums are unsigned, though,
so interpreting them as unsigned matches real world better.

Change btf_dump test case to test maximum 32-bit value, instead of negative
value.

Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20200303003233.3496043-3-andriin@fb.com
libbpf: Fix handling of optional field_name in btf_dump__emit_type_decl 2020-03-04T00:41:36+00:00 Andrii Nakryiko andriin@fb.com 2020-03-03T18:08:00+00:00 320a36063e1441210106aa33997ad3770d4c86b4 Internal functions, used by btf_dump__emit_type_decl(), assume field_name is never going to be NULL. Ensure it's always the case. Fixes: 9f81654eebe8 ("libbpf: Expose BTF-to-C type declaration emitting API") Signed-off-by: Andrii Nakryiko <andriin@fb.com> Signed-off-by: Alexei Starovoitov <ast@kernel.org> Link: https://lore.kernel.org/bpf/20200303180800.3303471-1-andriin@fb.com 
Internal functions, used by btf_dump__emit_type_decl(), assume field_name is
never going to be NULL. Ensure it's always the case.

Fixes: 9f81654eebe8 ("libbpf: Expose BTF-to-C type declaration emitting API")
Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200303180800.3303471-1-andriin@fb.com
libbpf: Fix error handling bug in btf_dump__new 2020-01-17T16:33:17+00:00 Andrii Nakryiko andriin@fb.com 2020-01-17T06:07:58+00:00 bc0eb9a333918d2cc4f42f7f4cf09eaf029ac9de Fix missing jump to error handling in btf_dump__new, found by Coverity static code analysis. Fixes: 9f81654eebe8 ("libbpf: Expose BTF-to-C type declaration emitting API") Signed-off-by: Andrii Nakryiko <andriin@fb.com> Signed-off-by: Alexei Starovoitov <ast@kernel.org> Link: https://lore.kernel.org/bpf/20200117060801.1311525-2-andriin@fb.com 
Fix missing jump to error handling in btf_dump__new, found by Coverity static
code analysis.

Fixes: 9f81654eebe8 ("libbpf: Expose BTF-to-C type declaration emitting API")
Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200117060801.1311525-2-andriin@fb.com