Message ID | 20200401110907.2669564-3-jolsa@kernel.org |
---|---|
State | RFC |
Delegated to: | BPF Maintainers |
Headers | show |
Series | bpf: Add d_path helper | expand |
On Wed, 2020-04-01 at 13:09 +0200, Jiri Olsa wrote: > + * int bpf_d_path(struct path *path, char *buf, u32 sz) > + * Description > + * Return full path for given 'struct path' object, which > + * needs to be the kernel BTF 'path' object. The path is > + * returned in buffer provided 'buf' of size 'sz'. > + * > + * Return > + * length of returned string on success, or a negative > + * error in case of failure > + * You might want to add that d_path is ambiguous since it can add " (deleted)" at the end of your path and you don't know whether this is actually part of the file path or not. :) > +BPF_CALL_3(bpf_d_path, struct path *, path, char *, buf, u32, sz) > +{ > + char *p = d_path(path, buf, sz - 1); I am curious why you'd use sz - 1 here? In my experience, d_path's output is 0 limited so you shouldn't need to keep an extra byte for that (if that was the intention here). > + int len; > + > + if (IS_ERR(p)) { > + len = PTR_ERR(p); > + } else { > + len = strlen(p); > + if (len && p != buf) { > + memmove(buf, p, len); Have you considered returning the offset within buf instead and let the BPF program do pointer arithmetics to find the beginning of the string? > + buf[len] = 0; If my previous comment about sz - 1 is true, then this wouldn't be necessary, you could just use memmove with len + 1. > + } > + } > + > + return len; > +}
On Thu, Apr 02, 2020 at 04:02:55PM +0200, Florent Revest wrote: > On Wed, 2020-04-01 at 13:09 +0200, Jiri Olsa wrote: > > + * int bpf_d_path(struct path *path, char *buf, u32 sz) > > + * Description > > + * Return full path for given 'struct path' object, which > > + * needs to be the kernel BTF 'path' object. The path is > > + * returned in buffer provided 'buf' of size 'sz'. > > + * > > + * Return > > + * length of returned string on success, or a negative > > + * error in case of failure > > + * > > You might want to add that d_path is ambiguous since it can add > " (deleted)" at the end of your path and you don't know whether this is > actually part of the file path or not. :) right > > > +BPF_CALL_3(bpf_d_path, struct path *, path, char *, buf, u32, sz) > > +{ > > + char *p = d_path(path, buf, sz - 1); > > I am curious why you'd use sz - 1 here? In my experience, d_path's > output is 0 limited so you shouldn't need to keep an extra byte for > that (if that was the intention here). > > > + int len; > > + > > + if (IS_ERR(p)) { > > + len = PTR_ERR(p); > > + } else { > > + len = strlen(p); > > + if (len && p != buf) { > > + memmove(buf, p, len); > > Have you considered returning the offset within buf instead and let the > BPF program do pointer arithmetics to find the beginning of the string? we could do that.. I was following some other user of d_path, which I can't find at the moment ;-) I'll check > > > + buf[len] = 0; > > If my previous comment about sz - 1 is true, then this wouldn't be > necessary, you could just use memmove with len + 1. hum, you might be right, I'll check on this thanks, jirka > > > + } > > + } > > + > > + return len; > > +} >
On Fri, Apr 3, 2020 at 2:03 AM Jiri Olsa <jolsa@redhat.com> wrote: > > On Thu, Apr 02, 2020 at 04:02:55PM +0200, Florent Revest wrote: > > On Wed, 2020-04-01 at 13:09 +0200, Jiri Olsa wrote: > > > + * int bpf_d_path(struct path *path, char *buf, u32 sz) > > > + * Description > > > + * Return full path for given 'struct path' object, which > > > + * needs to be the kernel BTF 'path' object. The path is > > > + * returned in buffer provided 'buf' of size 'sz'. > > > + * > > > + * Return > > > + * length of returned string on success, or a negative > > > + * error in case of failure > > > + * > > > > You might want to add that d_path is ambiguous since it can add > > " (deleted)" at the end of your path and you don't know whether this is > > actually part of the file path or not. :) > > right > > > > > > +BPF_CALL_3(bpf_d_path, struct path *, path, char *, buf, u32, sz) > > > +{ > > > + char *p = d_path(path, buf, sz - 1); > > > > I am curious why you'd use sz - 1 here? In my experience, d_path's > > output is 0 limited so you shouldn't need to keep an extra byte for > > that (if that was the intention here). > > > > > + int len; > > > + > > > + if (IS_ERR(p)) { > > > + len = PTR_ERR(p); > > > + } else { > > > + len = strlen(p); > > > + if (len && p != buf) { > > > + memmove(buf, p, len); > > > > Have you considered returning the offset within buf instead and let the > > BPF program do pointer arithmetics to find the beginning of the string? > > we could do that.. I was following some other user of d_path, > which I can't find at the moment ;-) I'll check This would make it hard to support variable-length data encoding and sending it over perf_buffer, because it would prevent back-to-back "stitching" of multiple strings compactly in output buffer. So I think current approach is preferable. > > > > > > + buf[len] = 0; > > > > If my previous comment about sz - 1 is true, then this wouldn't be > > necessary, you could just use memmove with len + 1. > > hum, you might be right, I'll check on this > > thanks, > jirka > > > > > > + } > > > + } > > > + > > > + return len; > > > +} > > >
diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h index 2e29a671d67e..8da1b4750364 100644 --- a/include/uapi/linux/bpf.h +++ b/include/uapi/linux/bpf.h @@ -3025,6 +3025,17 @@ union bpf_attr { * * **-EOPNOTSUPP** Unsupported operation, for example a * call from outside of TC ingress. * * **-ESOCKTNOSUPPORT** Socket type not supported (reuseport). + * + * int bpf_d_path(struct path *path, char *buf, u32 sz) + * Description + * Return full path for given 'struct path' object, which + * needs to be the kernel BTF 'path' object. The path is + * returned in buffer provided 'buf' of size 'sz'. + * + * Return + * length of returned string on success, or a negative + * error in case of failure + * */ #define __BPF_FUNC_MAPPER(FN) \ FN(unspec), \ @@ -3151,7 +3162,8 @@ union bpf_attr { FN(xdp_output), \ FN(get_netns_cookie), \ FN(get_current_ancestor_cgroup_id), \ - FN(sk_assign), + FN(sk_assign), \ + FN(d_path), /* integer value in 'imm' field of BPF_CALL instruction selects which helper * function eBPF program intends to call diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c index ca1796747a77..6ca390b2b26e 100644 --- a/kernel/trace/bpf_trace.c +++ b/kernel/trace/bpf_trace.c @@ -779,6 +779,35 @@ static const struct bpf_func_proto bpf_send_signal_thread_proto = { .arg1_type = ARG_ANYTHING, }; +BPF_CALL_3(bpf_d_path, struct path *, path, char *, buf, u32, sz) +{ + char *p = d_path(path, buf, sz - 1); + int len; + + if (IS_ERR(p)) { + len = PTR_ERR(p); + } else { + len = strlen(p); + if (len && p != buf) { + memmove(buf, p, len); + buf[len] = 0; + } + } + + return len; +} + +static u32 bpf_d_path_btf_ids[3]; +static const struct bpf_func_proto bpf_d_path_proto = { + .func = bpf_d_path, + .gpl_only = true, + .ret_type = RET_INTEGER, + .arg1_type = ARG_PTR_TO_BTF_ID, + .arg2_type = ARG_PTR_TO_MEM, + .arg3_type = ARG_CONST_SIZE, + .btf_id = bpf_d_path_btf_ids, +}; + const struct bpf_func_proto * bpf_tracing_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog) { @@ -1224,6 +1253,8 @@ tracing_prog_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog) case BPF_FUNC_xdp_output: return &bpf_xdp_output_proto; #endif + case BPF_FUNC_d_path: + return &bpf_d_path_proto; default: return raw_tp_prog_func_proto(func_id, prog); } diff --git a/scripts/bpf_helpers_doc.py b/scripts/bpf_helpers_doc.py index f43d193aff3a..8f62cbc4c3ff 100755 --- a/scripts/bpf_helpers_doc.py +++ b/scripts/bpf_helpers_doc.py @@ -418,6 +418,7 @@ class PrinterHelpers(Printer): 'struct __sk_buff', 'struct sk_msg_md', 'struct xdp_md', + 'struct path', ] known_types = { '...', @@ -450,6 +451,7 @@ class PrinterHelpers(Printer): 'struct sk_reuseport_md', 'struct sockaddr', 'struct tcphdr', + 'struct path', } mapped_types = { 'u8': '__u8', diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h index 2e29a671d67e..8da1b4750364 100644 --- a/tools/include/uapi/linux/bpf.h +++ b/tools/include/uapi/linux/bpf.h @@ -3025,6 +3025,17 @@ union bpf_attr { * * **-EOPNOTSUPP** Unsupported operation, for example a * call from outside of TC ingress. * * **-ESOCKTNOSUPPORT** Socket type not supported (reuseport). + * + * int bpf_d_path(struct path *path, char *buf, u32 sz) + * Description + * Return full path for given 'struct path' object, which + * needs to be the kernel BTF 'path' object. The path is + * returned in buffer provided 'buf' of size 'sz'. + * + * Return + * length of returned string on success, or a negative + * error in case of failure + * */ #define __BPF_FUNC_MAPPER(FN) \ FN(unspec), \ @@ -3151,7 +3162,8 @@ union bpf_attr { FN(xdp_output), \ FN(get_netns_cookie), \ FN(get_current_ancestor_cgroup_id), \ - FN(sk_assign), + FN(sk_assign), \ + FN(d_path), /* integer value in 'imm' field of BPF_CALL instruction selects which helper * function eBPF program intends to call
Adding d_path helper function that returns full path for give 'struct path' object, which needs to be the kernel BTF 'path' object. The helper calls directly d_path function. Updating also bpf.h tools uapi header and adding 'path' to bpf_helpers_doc.py script. Signed-off-by: Jiri Olsa <jolsa@kernel.org> --- include/uapi/linux/bpf.h | 14 +++++++++++++- kernel/trace/bpf_trace.c | 31 +++++++++++++++++++++++++++++++ scripts/bpf_helpers_doc.py | 2 ++ tools/include/uapi/linux/bpf.h | 14 +++++++++++++- 4 files changed, 59 insertions(+), 2 deletions(-)