| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <CAP-5=fWFUx7nKECSM4zD7bxYGiyopK5eXKRh-FYbzrM4SgrfjQ@mail.gmail.com>
Date: Wed, 6 Dec 2023 21:13:54 -0800
From: Ian Rogers <irogers@...gle.com>
To: Arnaldo Carvalho de Melo <acme@...nel.org>
Cc: Peter Zijlstra <peterz@...radead.org>,
Ingo Molnar <mingo@...hat.com>,
Mark Rutland <mark.rutland@....com>,
Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
Jiri Olsa <jolsa@...nel.org>,
Namhyung Kim <namhyung@...nel.org>,
Adrian Hunter <adrian.hunter@...el.com>,
Nick Terrell <terrelln@...com>,
Kan Liang <kan.liang@...ux.intel.com>,
Andi Kleen <ak@...ux.intel.com>,
Kajol Jain <kjain@...ux.ibm.com>,
Athira Rajeev <atrajeev@...ux.vnet.ibm.com>,
Huacai Chen <chenhuacai@...nel.org>,
Masami Hiramatsu <mhiramat@...nel.org>,
Vincent Whitchurch <vincent.whitchurch@...s.com>,
"Steinar H. Gunderson" <sesse@...gle.com>,
Liam Howlett <liam.howlett@...cle.com>,
Miguel Ojeda <ojeda@...nel.org>,
Colin Ian King <colin.i.king@...il.com>,
Dmitrii Dolgov <9erthalion6@...il.com>,
Yang Jihong <yangjihong1@...wei.com>,
Ming Wang <wangming01@...ngson.cn>,
James Clark <james.clark@....com>,
K Prateek Nayak <kprateek.nayak@....com>,
Sean Christopherson <seanjc@...gle.com>,
Leo Yan <leo.yan@...aro.org>,
Ravi Bangoria <ravi.bangoria@....com>,
German Gomez <german.gomez@....com>,
Changbin Du <changbin.du@...wei.com>,
Paolo Bonzini <pbonzini@...hat.com>, Li Dong <lidong@...o.com>,
Sandipan Das <sandipan.das@....com>,
liuwenyu <liuwenyu7@...wei.com>, linux-kernel@...r.kernel.org,
linux-perf-users@...r.kernel.org,
Guilherme Amadio <amadio@...too.org>
Subject: Re: [PATCH v5 06/50] tools lib api: Add io_dir an allocation free
readdir alternative
On Thu, Nov 30, 2023 at 1:25 PM Arnaldo Carvalho de Melo
<acme@...nel.org> wrote:
>
> Em Thu, Nov 30, 2023 at 09:56:42AM -0800, Ian Rogers escreveu:
> > On Thu, Nov 30, 2023 at 9:21 AM Arnaldo Carvalho de Melo
> > <acme@...nel.org> wrote:
> > >
> > > Em Mon, Nov 27, 2023 at 02:08:18PM -0800, Ian Rogers escreveu:
> > > > glibc's opendir allocates a minimum of 32kb, when called recursively
> > > > for a directory tree the memory consumption can add up - nearly 300kb
> > > > during perf start-up when processing modules. Add a stack allocated
> > > > variant of readdir sized a little more than 1kb.
> > >
> > > Now, on some systems:
> > >
> > > CC /tmp/build/perf/pmu-events/pmu-events.o
> > > In file included from util/machine.c:39:
> > > /tmp/build/perf/libapi/include/api/io_dir.h: In function ‘io_dir__readdir’:
> > > /tmp/build/perf/libapi/include/api/io_dir.h:46:16: error: implicit declaration of function ‘getdents64’; did you mean ‘getentropy’? [-Werror=implicit-function-declaration]
> > > ssize_t rc = getdents64(iod->dirfd, iod->buff, sizeof(iod->buff));
> > > ^~~~~~~~~~
> > > getentropy
> > > CC /tmp/build/perf/tests/workloads/brstack.o
> > > CC /tmp/build/perf/tests/workloads/datasym.o
> > > CC /tmp/build/perf/util/maps.o
> >
> >
> > Sorry for that. Is it a _GNU_SOURCE issue? I thought we generally had
>
> Nope, some systems just don't have that function in its libc, for
> instance, a Red Hat derivative with glibc 2.28:
>
> glibc-2.28-189.5.el8_6.x86_64
>
> [perfbuilder@...45cef2169 /]$ grep -r getdents64 /usr/include
> /usr/include/bits/syscall.h:#ifdef __NR_getdents64
> /usr/include/bits/syscall.h:# define SYS_getdents64 __NR_getdents64
> /usr/include/asm-generic/unistd.h:#define __NR_getdents64 61
> /usr/include/asm-generic/unistd.h:__SYSCALL(__NR_getdents64, sys_getdents64)
> /usr/include/sanitizer/linux_syscall_hooks.h:#define __sanitizer_syscall_pre_getdents64(fd, dirent, count) \
> /usr/include/sanitizer/linux_syscall_hooks.h: __sanitizer_syscall_pre_impl_getdents64((long)(fd), (long)(dirent), \
> /usr/include/sanitizer/linux_syscall_hooks.h:#define __sanitizer_syscall_post_getdents64(res, fd, dirent, count) \
> /usr/include/sanitizer/linux_syscall_hooks.h: __sanitizer_syscall_post_impl_getdents64(res, (long)(fd), (long)(dirent), \
> /usr/include/sanitizer/linux_syscall_hooks.h:void __sanitizer_syscall_pre_impl_getdents64(long fd, long dirent, long count);
> /usr/include/sanitizer/linux_syscall_hooks.h:void __sanitizer_syscall_post_impl_getdents64(long res, long fd, long dirent,
> /usr/include/asm/unistd_64.h:#define __NR_getdents64 217
> /usr/include/asm/unistd_x32.h:#define __NR_getdents64 (__X32_SYSCALL_BIT + 217)
> /usr/include/asm/unistd_32.h:#define __NR_getdents64 220
> [perfbuilder@...45cef2169 /]$
>
> Systems with musl libc fail differently:
>
> 6 19.51 alpine:3.18 : FAIL gcc version 12.2.1 20220924 (Alpine 12.2.1_git20220924-r10)
> CC /tmp/build/perf/util/sample-raw.o
> CC /tmp/build/perf/tests/kmod-path.o
> CC /tmp/build/perf/tests/thread-map.o
> CC /tmp/build/perf/tests/topology.o
> In file included from util/machine.c:39:
> /tmp/build/perf/libapi/include/api/io_dir.h: In function 'io_dir__readdir':
> /tmp/build/perf/libapi/include/api/io_dir.h:46:56: error: passing argument 2 of 'getdents' from incompatible pointer type [-Werror=incompatible-pointer-types]
> 46 | ssize_t rc = getdents64(iod->dirfd, iod->buff, sizeof(iod->buff));
> | ~~~^~~~~~
> | |
> | struct io_dirent64 *
> In file included from util/machine.c:2:
> /usr/include/dirent.h:52:19: note: expected 'struct dirent *' but argument is of type 'struct io_dirent64 *'
> 52 | int getdents(int, struct dirent *, size_t);
> | ^~~~~~~~~~~~~~~
>
> Their devel distro also fails like that:
>
> 7 20.49 alpine:edge : FAIL gcc version 13.1.1 20230722 (Alpine 13.1.1_git20230722)
> CC /tmp/build/perf/tests/sdt.o
> CC /tmp/build/perf/util/maps.o
> CC /tmp/build/perf/tests/is_printable_array.o
> CC /tmp/build/perf/util/intel-pt-decoder/intel-pt-insn-decoder.o
> In file included from util/machine.c:39:
> /tmp/build/perf/libapi/include/api/io_dir.h: In function 'io_dir__readdir':
> /tmp/build/perf/libapi/include/api/io_dir.h:46:56: error: passing argument 2 of 'getdents' from incompatible pointer type [-Werror=incompatible-pointer-types]
> 46 | ssize_t rc = getdents64(iod->dirfd, iod->buff, sizeof(iod->buff));
> | ~~~^~~~~~
> | |
> | struct io_dirent64 *
> In file included from util/machine.c:2:
> /usr/include/dirent.h:52:19: note: expected 'struct dirent *' but argument is of type 'struct io_dirent64 *'
> 52 | int getdents(int, struct dirent *, size_t);
> | ^~~~~~~~~~~~~~~
> CC /tmp/build/perf/tests/bitmap.o
>
> And then we have nolibc:
>
> ⬢[acme@...lbox perf-tools-next]$ find tools/ -name nolibc
> tools/include/nolibc
> tools/testing/selftests/nolibc
> ⬢[acme@...lbox perf-tools-next]$ grep getdents64 tools/include/nolibc/*
> tools/include/nolibc/sys.h: * int getdents64(int fd, struct linux_dirent64 *dirp, int count);
> tools/include/nolibc/sys.h:int sys_getdents64(int fd, struct linux_dirent64 *dirp, int count)
> tools/include/nolibc/sys.h: return my_syscall3(__NR_getdents64, fd, dirp, count);
> tools/include/nolibc/sys.h:int getdents64(int fd, struct linux_dirent64 *dirp, int count)
> tools/include/nolibc/sys.h: return __sysret(sys_getdents64(fd, dirp, count));
> tools/include/nolibc/types.h:/* for getdents64() */
> ⬢[acme@...lbox perf-tools-next]$
>
> On musl libc systems:
>
> / $ grep -A1 -B20 getdents64 /usr/include/*.h
> /usr/include/dirent.h-#define DT_SOCK 12
> /usr/include/dirent.h-#define DT_WHT 14
> /usr/include/dirent.h-#define IFTODT(x) ((x)>>12 & 017)
> /usr/include/dirent.h-#define DTTOIF(x) ((x)<<12)
> /usr/include/dirent.h-int getdents(int, struct dirent *, size_t);
> /usr/include/dirent.h-#endif
> /usr/include/dirent.h-
> /usr/include/dirent.h-#ifdef _GNU_SOURCE
> /usr/include/dirent.h-int versionsort(const struct dirent **, const struct dirent **);
> /usr/include/dirent.h-#endif
> /usr/include/dirent.h-
> /usr/include/dirent.h-#if defined(_LARGEFILE64_SOURCE) || defined(_GNU_SOURCE)
> /usr/include/dirent.h-#define dirent64 dirent
> /usr/include/dirent.h-#define readdir64 readdir
> /usr/include/dirent.h-#define readdir64_r readdir_r
> /usr/include/dirent.h-#define scandir64 scandir
> /usr/include/dirent.h-#define alphasort64 alphasort
> /usr/include/dirent.h-#define versionsort64 versionsort
> /usr/include/dirent.h-#define off64_t off_t
> /usr/include/dirent.h-#define ino64_t ino_t
> /usr/include/dirent.h:#define getdents64 getdents
> /usr/include/dirent.h-#endif
> / $
I'm confused. It seems nolibc has support, musl doesn't and with old
glibcs it depends. I added a feature test in a the latest patch series
to hopefully workaround this:
https://lore.kernel.org/lkml/20231207050433.1426834-1-irogers@google.com/
Could you help check and I suspect the syscall implementation needs
some work for if SYS_getdents64 isn't defined, etc.
Thanks,
Ian
> > _GNU_SOURCE defined on the command line for reallocarray. Maybe we
> > shouldn't define this on the command line and do it immediately before
> > the relevant include with:
> >
> > #ifndef _GNU_SOURCE
> > #define _GNU_SOURCE
> > #endif
> >
> > like in tools/perf/util/intel-pt-decoder/intel-pt-decoder.c
> >
> > Thanks,
> > Ian
> >
> > >
> > > > Signed-off-by: Ian Rogers <irogers@...gle.com>
> > > > ---
> > > > tools/lib/api/Makefile | 2 +-
> > > > tools/lib/api/io_dir.h | 75 ++++++++++++++++++++++++++++++++++++++++++
> > > > 2 files changed, 76 insertions(+), 1 deletion(-)
> > > > create mode 100644 tools/lib/api/io_dir.h
> > > >
> > > > diff --git a/tools/lib/api/Makefile b/tools/lib/api/Makefile
> > > > index 044860ac1ed1..186aa407de8c 100644
> > > > --- a/tools/lib/api/Makefile
> > > > +++ b/tools/lib/api/Makefile
> > > > @@ -99,7 +99,7 @@ install_lib: $(LIBFILE)
> > > > $(call do_install_mkdir,$(libdir_SQ)); \
> > > > cp -fpR $(LIBFILE) $(DESTDIR)$(libdir_SQ)
> > > >
> > > > -HDRS := cpu.h debug.h io.h
> > > > +HDRS := cpu.h debug.h io.h io_dir.h
> > > > FD_HDRS := fd/array.h
> > > > FS_HDRS := fs/fs.h fs/tracing_path.h
> > > > INSTALL_HDRS_PFX := $(DESTDIR)$(prefix)/include/api
> > > > diff --git a/tools/lib/api/io_dir.h b/tools/lib/api/io_dir.h
> > > > new file mode 100644
> > > > index 000000000000..f3479006edb6
> > > > --- /dev/null
> > > > +++ b/tools/lib/api/io_dir.h
> > > > @@ -0,0 +1,75 @@
> > > > +/* SPDX-License-Identifier: (LGPL-2.1 OR BSD-2-Clause) */
> > > > +/*
> > > > + * Lightweight directory reading library.
> > > > + */
> > > > +#ifndef __API_IO_DIR__
> > > > +#define __API_IO_DIR__
> > > > +
> > > > +#include <dirent.h>
> > > > +#include <fcntl.h>
> > > > +#include <stdlib.h>
> > > > +#include <unistd.h>
> > > > +#include <sys/stat.h>
> > > > +
> > > > +struct io_dirent64 {
> > > > + ino64_t d_ino; /* 64-bit inode number */
> > > > + off64_t d_off; /* 64-bit offset to next structure */
> > > > + unsigned short d_reclen; /* Size of this dirent */
> > > > + unsigned char d_type; /* File type */
> > > > + char d_name[NAME_MAX + 1]; /* Filename (null-terminated) */
> > > > +};
> > > > +
> > > > +struct io_dir {
> > > > + int dirfd;
> > > > + ssize_t available_bytes;
> > > > + struct io_dirent64 *next;
> > > > + struct io_dirent64 buff[4];
> > > > +};
> > > > +
> > > > +static inline void io_dir__init(struct io_dir *iod, int dirfd)
> > > > +{
> > > > + iod->dirfd = dirfd;
> > > > + iod->available_bytes = 0;
> > > > +}
> > > > +
> > > > +static inline void io_dir__rewinddir(struct io_dir *iod)
> > > > +{
> > > > + lseek(iod->dirfd, 0, SEEK_SET);
> > > > + iod->available_bytes = 0;
> > > > +}
> > > > +
> > > > +static inline struct io_dirent64 *io_dir__readdir(struct io_dir *iod)
> > > > +{
> > > > + struct io_dirent64 *entry;
> > > > +
> > > > + if (iod->available_bytes <= 0) {
> > > > + ssize_t rc = getdents64(iod->dirfd, iod->buff, sizeof(iod->buff));
> > > > +
> > > > + if (rc <= 0)
> > > > + return NULL;
> > > > + iod->available_bytes = rc;
> > > > + iod->next = iod->buff;
> > > > + }
> > > > + entry = iod->next;
> > > > + iod->next = (struct io_dirent64 *)((char *)entry + entry->d_reclen);
> > > > + iod->available_bytes -= entry->d_reclen;
> > > > + return entry;
> > > > +}
> > > > +
> > > > +static inline bool io_dir__is_dir(const struct io_dir *iod, struct io_dirent64 *dent)
> > > > +{
> > > > + if (dent->d_type == DT_UNKNOWN) {
> > > > + struct stat st;
> > > > +
> > > > + if (fstatat(iod->dirfd, dent->d_name, &st, /*flags=*/0))
> > > > + return false;
> > > > +
> > > > + if (S_ISDIR(st.st_mode)) {
> > > > + dent->d_type = DT_DIR;
> > > > + return true;
> > > > + }
> > > > + }
> > > > + return dent->d_type == DT_DIR;
> > > > +}
> > > > +
> > > > +#endif
> > > > --
> > > > 2.43.0.rc1.413.gea7ed67945-goog
> > > >
> > >
> > > --
> > >
> > > - Arnaldo
>
> --
>
> - Arnaldo
Powered by blists - more mailing lists