linux-stable.git/tools/bpf, branch v6.2.3

bpftool: Always disable stack protection for BPF objects

2023-03-10T08:29:08+00:00

[ Upstream commit 878625e1c7a10dfbb1fdaaaae2c4d2a58fbce627 ]

When the clang toolchain has stack protection enabled in order to be
consistent with gcc - which just happens to be the case on Gentoo -
the bpftool build fails:

  [...]
  clang \
	-I. \
	-I/tmp/portage/dev-util/bpftool-6.0.12/work/linux-6.0/tools/include/uapi/ \
	-I/tmp/portage/dev-util/bpftool-6.0.12/work/linux-6.0/tools/bpf/bpftool/bootstrap/libbpf/include \
	-g -O2 -Wall -target bpf -c skeleton/pid_iter.bpf.c -o pid_iter.bpf.o
  clang \
	-I. \
	-I/tmp/portage/dev-util/bpftool-6.0.12/work/linux-6.0/tools/include/uapi/ \
	-I/tmp/portage/dev-util/bpftool-6.0.12/work/linux-6.0/tools/bpf/bpftool/bootstrap/libbpf/include \
	-g -O2 -Wall -target bpf -c skeleton/profiler.bpf.c -o profiler.bpf.o
  skeleton/profiler.bpf.c:40:14: error: A call to built-in function '__stack_chk_fail' is not supported.
  int BPF_PROG(fentry_XXX)
                ^
  skeleton/profiler.bpf.c:94:14: error: A call to built-in function '__stack_chk_fail' is not supported.
  int BPF_PROG(fexit_XXX)
                ^
  2 errors generated.
  [...]

Since stack-protector makes no sense for the BPF bits just unconditionally
disable it.

Bug: https://bugs.gentoo.org/890638
Signed-off-by: Holger Hoffstätte 
Signed-off-by: Daniel Borkmann 
Acked-by: Quentin Monnet 
Link: https://lore.kernel.org/bpf/74cd9d2e-6052-312a-241e-2b514a75c92c@applied-asynchrony.com
Signed-off-by: Sasha Levin

bpftool: profile online CPUs instead of possible

2023-03-10T08:28:12+00:00

[ Upstream commit 377c16fa3f3c60d21e4b05314c8be034ce37f2eb ]

The number of online cpu may be not equal to possible cpu.
"bpftool prog profile" can not create pmu event on possible
but on online cpu.

$ dmidecode -s system-product-name
PowerEdge R620
$ cat /sys/devices/system/cpu/possible
0-47
$ cat /sys/devices/system/cpu/online
0-31

Disable cpu dynamically:
$ echo 0 > /sys/devices/system/cpu/cpuX/online

If one cpu is offline, perf_event_open will return ENODEV.
To fix this issue:
* check value returned and skip offline cpu.
* close pmu_fd immediately on error path, avoid fd leaking.

Fixes: 47c09d6a9f67 ("bpftool: Introduce "prog profile" command")
Signed-off-by: Tonghao Zhang 
Cc: Quentin Monnet 
Cc: Alexei Starovoitov 
Cc: Daniel Borkmann 
Cc: Andrii Nakryiko 
Cc: Martin KaFai Lau 
Cc: Song Liu 
Cc: Yonghong Song 
Cc: John Fastabend 
Cc: KP Singh 
Cc: Stanislav Fomichev 
Cc: Hao Luo 
Cc: Jiri Olsa 
Acked-by: John Fastabend 
Link: https://lore.kernel.org/r/20230202131701.29519-1-tong@infragraf.org
Signed-off-by: Martin KaFai Lau 
Signed-off-by: Sasha Levin

bpftool: Fix linkage with statically built libllvm

2022-12-22T19:09:43+00:00

Since the commit eb9d1acf634b ("bpftool: Add LLVM as default library for
disassembling JIT-ed programs") we might link the bpftool program with the
libllvm library. This works fine when a shared libllvm library is available,
but fails if we want to link bpftool with a statically built LLVM:

  [...]
  /usr/bin/ld: /usr/local/lib/libLLVMSupport.a(CrashRecoveryContext.cpp.o): in function `llvm::CrashRecoveryContextCleanup::~CrashRecoveryContextCleanup()':
  CrashRecoveryContext.cpp:(.text._ZN4llvm27CrashRecoveryContextCleanupD0Ev+0x17): undefined reference to `operator delete(void*, unsigned long)'
  /usr/bin/ld: /usr/local/lib/libLLVMSupport.a(CrashRecoveryContext.cpp.o): in function `llvm::CrashRecoveryContext::~CrashRecoveryContext()':
  CrashRecoveryContext.cpp:(.text._ZN4llvm20CrashRecoveryContextD2Ev+0xc8): undefined reference to `operator delete(void*, unsigned long)'
  [...]

So in the case of static libllvm we need to explicitly link bpftool with
required libraries, namely, libstdc++ and those provided by the `llvm-config
--system-libs` command. We can distinguish between the shared and static cases
by using the `llvm-config --shared-mode` command.

Fixes: eb9d1acf634b ("bpftool: Add LLVM as default library for disassembling JIT-ed programs")
Signed-off-by: Anton Protopopov 
Signed-off-by: Daniel Borkmann 
Acked-by: Stanislav Fomichev 
Link: https://lore.kernel.org/bpf/20221222102627.1643709-1-aspsk@isovalent.com

bpftool: Fix memory leak in do_build_table_cb

2022-12-06T20:20:42+00:00

strdup() allocates memory for path. We need to release the memory in the
following error path. Add free() to avoid memory leak.

Fixes: 8f184732b60b ("bpftool: Switch to libbpf's hashmap for pinned paths of BPF objects")
Signed-off-by: Miaoqian Lin 
Signed-off-by: Daniel Borkmann 
Link: https://lore.kernel.org/bpf/20221206071906.806384-1-linmq006@gmail.com

bpftool: remove function free_btf_vmlinux()

2022-11-21T00:17:46+00:00

The function contains a single btf__free() call which can be
inlined. Credits to Yonghong Song.

Signed-off-by: Sahid Orentino Ferdjaoui 
Acked-by: Yonghong Song 
Suggested-by: Yonghong Song 
Reviewed-by: Quentin Monnet 
Link: https://lore.kernel.org/r/20221120112515.38165-6-sahid.ferdjaoui@industrialdiscipline.com
Signed-off-by: Alexei Starovoitov

bpftool: clean-up usage of libbpf_get_error()

2022-11-21T00:17:46+00:00

bpftool is now totally compliant with libbpf 1.0 mode and is not
expected to be compiled with pre-1.0, let's clean-up the usage of
libbpf_get_error().

The changes stay aligned with returned errors always negative.

- In tools/bpf/bpftool/btf.c This fixes an uninitialized local
variable `err` in function do_dump() because it may now be returned
without having been set.
- This also removes the checks on NULL pointers before calling
btf__free() because that function already does the check.

Signed-off-by: Sahid Orentino Ferdjaoui 
Link: https://lore.kernel.org/r/20221120112515.38165-5-sahid.ferdjaoui@industrialdiscipline.com
Signed-off-by: Alexei Starovoitov

bpftool: fix error message when function can't register struct_ops

2022-11-21T00:17:46+00:00

It is expected that errno be passed to strerror(). This also cleans
this part of code from using libbpf_get_error().

Signed-off-by: Sahid Orentino Ferdjaoui 
Acked-by: Yonghong Song 
Suggested-by: Quentin Monnet 
Reviewed-by: Quentin Monnet 
Link: https://lore.kernel.org/r/20221120112515.38165-4-sahid.ferdjaoui@industrialdiscipline.com
Signed-off-by: Alexei Starovoitov

bpftool: replace return value PTR_ERR(NULL) with 0

2022-11-21T00:17:46+00:00

There is no reasons to keep PTR_ERR() when kern_btf=NULL, let's just
return 0.
This also cleans this part of code from using libbpf_get_error().

Signed-off-by: Sahid Orentino Ferdjaoui 
Acked-by: Yonghong Song 
Suggested-by: Quentin Monnet 
Link: https://lore.kernel.org/r/20221120112515.38165-3-sahid.ferdjaoui@industrialdiscipline.com
Signed-off-by: Alexei Starovoitov

bpftool: remove support of --legacy option for bpftool

2022-11-21T00:17:46+00:00

Following:
  commit bd054102a8c7 ("libbpf: enforce strict libbpf 1.0 behaviors")
  commit 93b8952d223a ("libbpf: deprecate legacy BPF map definitions")

The --legacy option is no longer relevant as libbpf no longer supports
it. libbpf_set_strict_mode() is a no-op operation.

Signed-off-by: Sahid Orentino Ferdjaoui 
Acked-by: Yonghong Song 
Reviewed-by: Quentin Monnet 
Link: https://lore.kernel.org/r/20221120112515.38165-2-sahid.ferdjaoui@industrialdiscipline.com
Signed-off-by: Alexei Starovoitov

bpftool: Check argc first before "file" in do_batch()

2022-11-16T05:36:24+00:00

If the parameters for batch are more than 2, check argc first can
return immediately, no need to use is_prefix() to check "file" with
a little overhead and then check argc, it is better to check "file"
only when the parameters for batch are 2.

Signed-off-by: Tiezhu Yang 
Acked-by: Stanislav Fomichev 
Reviewed-by: Quentin Monnet 
Link: https://lore.kernel.org/r/1668517207-11822-1-git-send-email-yangtiezhu@loongson.cn
Signed-off-by: Alexei Starovoitov