- Linux-kselftest-mirror - lists.linaro.org

[Linux-kselftest-mirror] Difficulties for compilation without extra optimisation

by SF Markus Elfring

Hello, I came along some software modules where I suggested source code adjustments. Example: nfs/write: Use common error handling code in nfs_lock_and_join_requests() https://lkml.org/lkml/2017/11/7/599 https://patchwork.kernel.org/patch/10047013/ https://lkml.kernel.org/r/<7f072f78-eef4-6d87-d233-cee71dac5a32(a)users.sourceforge.net> I would like to check corresponding build results then without extra optimisation applied by the compiler. But I got surprised by error messages for a command like the following. elfring@Sonne:~/Projekte/Linux/next-patched> my_cc=/usr/bin/gcc-7 && LANG=C make -j4 CC="${my_cc}" HOSTCC="${my_cc}" EXTRA_CFLAGS='-O0' allmodconfig fs/nfs/write.o … In file included from ./include/linux/compiler.h:58:0, from ./include/uapi/linux/stddef.h:1, from ./include/linux/stddef.h:4, from ./include/uapi/linux/posix_types.h:4, from ./include/uapi/linux/types.h:13, from ./include/linux/types.h:5, from fs/nfs/write.c:9: ./arch/x86/include/asm/jump_label.h: In function ‘trace_nfs_writeback_page_enter’: ./include/linux/compiler-gcc.h:275:38: warning: asm operand 0 probably doesn’t match constraints #define asm_volatile_goto(x...) do { asm goto(x); asm (""); } while (0) … How do you think about to improve this software situation anyhow? Regards, Markus -- To unsubscribe from this list: send the line "unsubscribe linux-kselftest" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html

7 years, 6 months

3
8
0 0

[Linux-kselftest-mirror] [PATCH 0/5] bpf: correct broken uapi for BPF_PROG_TYPE_PERF_EVENT program type

by Hendrik Brueckner

Perf tool bpf selftests revealed a broken uapi for s390 and arm64. With the BPF_PROG_TYPE_PERF_EVENT program type the bpf_perf_event structure exports the pt_regs structure for all architectures. This fails for s390 and arm64 because pt_regs are not part of the user api and kept in-kernel only. To mitigate the broken uapi, introduce a wrapper that exports pt_regs in an asm-generic way. For arm64, export the exising user_pt_regs structure. For s390, introduce a user_pt_regs structure that exports the beginning of pt_regs. Note that user_pt_regs must export from the beginning of pt_regs as BPF_PROG_TYPE_PERF_EVENT program type is not the only type for running BPF programs. Some more background: For the bpf_perf_event, there is a uapi definition that is passed to the BPF program. For other "probe" points like trace points, kprobes, and uprobes, there is no uapi and the BPF program is always passed pt_regs (which is OK as the BPF program runs in the kernel context). The perf tool can attach BPF programs to all of these "probe" points and, optionally, can create a BPF prologue to access particular arguments (passed as registers). For this, it uses DWARF/CFI information to obtain the register and calls a perf-arch backend function, regs_query_register_offset(). This function returns the index into (user_)pt_regs for a particular register. Then, perf creates a BPF prologue that accesses this register based on the passed stucture from the "probe" point. Part of this series, are also updates to the testing and bpf selftest to deal with asm-specifics. To complete the bpf support in perf, the the regs_query_register_offset function is added for s390 to support BPF prologue creation. Hendrik Brueckner (5): bpf: correct broken uapi for BPF_PROG_TYPE_PERF_EVENT program type s390/bpf: correct broken uapi for BPF_PROG_TYPE_PERF_EVENT program type arm64/bpf: correct broken uapi for BPF_PROG_TYPE_PERF_EVENT program type selftests/bpf: sync kernel headers and introduce arch support in Makefile perf s390: add regs_query_register_offset() arch/arm64/include/asm/perf_event.h | 2 + arch/arm64/include/uapi/asm/bpf_perf_event.h | 9 + arch/s390/include/asm/perf_event.h | 1 + arch/s390/include/asm/ptrace.h | 11 +- arch/s390/include/uapi/asm/bpf_perf_event.h | 9 + arch/s390/include/uapi/asm/ptrace.h | 11 + include/linux/perf_event.h | 6 +- include/uapi/asm-generic/bpf_perf_event.h | 9 + include/uapi/linux/bpf_perf_event.h | 5 +- kernel/events/core.c | 2 +- tools/arch/arm64/include/uapi/asm/bpf_perf_event.h | 9 + tools/arch/s390/include/uapi/asm/bpf_perf_event.h | 9 + tools/arch/s390/include/uapi/asm/ptrace.h | 471 +++++++++++++++++++++ tools/include/uapi/asm-generic/bpf_perf_event.h | 9 + tools/include/uapi/linux/bpf_perf_event.h | 6 +- tools/perf/arch/s390/Makefile | 1 + tools/perf/arch/s390/util/dwarf-regs.c | 32 +- tools/perf/check-headers.sh | 1 + tools/testing/selftests/bpf/Makefile | 14 +- 19 files changed, 602 insertions(+), 15 deletions(-) create mode 100644 arch/arm64/include/uapi/asm/bpf_perf_event.h create mode 100644 arch/s390/include/uapi/asm/bpf_perf_event.h create mode 100644 include/uapi/asm-generic/bpf_perf_event.h create mode 100644 tools/arch/arm64/include/uapi/asm/bpf_perf_event.h create mode 100644 tools/arch/s390/include/uapi/asm/bpf_perf_event.h create mode 100644 tools/arch/s390/include/uapi/asm/ptrace.h create mode 100644 tools/include/uapi/asm-generic/bpf_perf_event.h -- 1.8.3.1 -- To unsubscribe from this list: send the line "unsubscribe linux-kselftest" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html

7 years, 6 months

2
6
0 0

[Linux-kselftest-mirror] selftests: Testing a write attempt into a full file?

by SF Markus Elfring

Hello, I have constructed another demonstration program. #include <errno.h> #include <stdio.h> #include <stdlib.h> int main(void) { FILE *f = fopen("/dev/full", "a"); if (!f) goto report_failure; { int const c = 'X'; if (fputc(c, f) != c) goto report_failure; } return EXIT_SUCCESS; report_failure: perror(__func__); return errno; } I got the following result. elfring@Sonne:~/Projekte/selftests> gcc-7 putc_into_full_file1.c && ./a.out; echo $? 0 Does such a simple test example need further software development considerations? Regards, Markus -- To unsubscribe from this list: send the line "unsubscribe linux-kselftest" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html

7 years, 6 months

2
1
0 0

[Linux-kselftest-mirror] [PATCH AUTOSEL for 4.9 04/52] x86/selftests: Add clobbers for int80 on x86_64

by alexander.levin＠verizon.com

From: Dmitry Safonov <dsafonov(a)virtuozzo.com> [ Upstream commit 2a4d0c627f5374f365a873dea4e10ae0bb437680 ] Kernel erases R8..R11 registers prior returning to userspace from int80: https://lkml.org/lkml/2009/10/1/164 GCC can reuse these registers and doesn't expect them to change during syscall invocation. I met this kind of bug in CRIU once GCC 6.1 and CLANG stored local variables in those registers and the kernel zerofied them during syscall: https://github.com/xemul/criu/commit/990d33f1a1cdd17bca6c2eb059ab3be2564f7f… By that reason I suggest to add those registers to clobbers in selftests. Also, as noted by Andy - removed unneeded clobber for flags in INT $0x80 inline asm. Signed-off-by: Dmitry Safonov <dsafonov(a)virtuozzo.com> Acked-by: Andy Lutomirski <luto(a)kernel.org> Cc: 0x7f454c46(a)gmail.com Cc: Borislav Petkov <bp(a)alien8.de> Cc: Borislav Petkov <bp(a)suse.de> Cc: Brian Gerst <brgerst(a)gmail.com> Cc: Denys Vlasenko <dvlasenk(a)redhat.com> Cc: H. Peter Anvin <hpa(a)zytor.com> Cc: Josh Poimboeuf <jpoimboe(a)redhat.com> Cc: Linus Torvalds <torvalds(a)linux-foundation.org> Cc: Peter Zijlstra <peterz(a)infradead.org> Cc: Shuah Khan <shuah(a)kernel.org> Cc: Thomas Gleixner <tglx(a)linutronix.de> Cc: linux-kselftest(a)vger.kernel.org Link: http://lkml.kernel.org/r/20170213101336.20486-1-dsafonov@virtuozzo.com Signed-off-by: Ingo Molnar <mingo(a)kernel.org> Signed-off-by: Sasha Levin <alexander.levin(a)verizon.com> --- tools/testing/selftests/x86/fsgsbase.c | 2 +- tools/testing/selftests/x86/ldt_gdt.c | 16 +++++++++++----- tools/testing/selftests/x86/ptrace_syscall.c | 3 ++- tools/testing/selftests/x86/single_step_syscall.c | 5 ++++- 4 files changed, 18 insertions(+), 8 deletions(-) diff --git a/tools/testing/selftests/x86/fsgsbase.c b/tools/testing/selftests/x86/fsgsbase.c index 9b4610c6d3fb..f249e042b3b5 100644 --- a/tools/testing/selftests/x86/fsgsbase.c +++ b/tools/testing/selftests/x86/fsgsbase.c @@ -245,7 +245,7 @@ void do_unexpected_base(void) long ret; asm volatile ("int $0x80" : "=a" (ret) : "a" (243), "b" (low_desc) - : "flags"); + : "r8", "r9", "r10", "r11"); memcpy(&desc, low_desc, sizeof(desc)); munmap(low_desc, sizeof(desc)); diff --git a/tools/testing/selftests/x86/ldt_gdt.c b/tools/testing/selftests/x86/ldt_gdt.c index e717fed80219..b9a22f18566a 100644 --- a/tools/testing/selftests/x86/ldt_gdt.c +++ b/tools/testing/selftests/x86/ldt_gdt.c @@ -45,6 +45,12 @@ #define AR_DB (1 << 22) #define AR_G (1 << 23) +#ifdef __x86_64__ +# define INT80_CLOBBERS "r8", "r9", "r10", "r11" +#else +# define INT80_CLOBBERS +#endif + static int nerrs; /* Points to an array of 1024 ints, each holding its own index. */ @@ -634,7 +640,7 @@ static int invoke_set_thread_area(void) asm volatile ("int $0x80" : "=a" (ret), "+m" (low_user_desc) : "a" (243), "b" (low_user_desc) - : "flags"); + : INT80_CLOBBERS); return ret; } @@ -703,7 +709,7 @@ static void test_gdt_invalidation(void) "+a" (eax) : "m" (low_user_desc_clear), [arg1] "r" ((unsigned int)(unsigned long)low_user_desc_clear) - : "flags"); + : INT80_CLOBBERS); if (sel != 0) { result = "FAIL"; @@ -734,7 +740,7 @@ static void test_gdt_invalidation(void) "+a" (eax) : "m" (low_user_desc_clear), [arg1] "r" ((unsigned int)(unsigned long)low_user_desc_clear) - : "flags"); + : INT80_CLOBBERS); if (sel != 0) { result = "FAIL"; @@ -767,7 +773,7 @@ static void test_gdt_invalidation(void) "+a" (eax) : "m" (low_user_desc_clear), [arg1] "r" ((unsigned int)(unsigned long)low_user_desc_clear) - : "flags"); + : INT80_CLOBBERS); #ifdef __x86_64__ syscall(SYS_arch_prctl, ARCH_GET_FS, &new_base); @@ -820,7 +826,7 @@ static void test_gdt_invalidation(void) "+a" (eax) : "m" (low_user_desc_clear), [arg1] "r" ((unsigned int)(unsigned long)low_user_desc_clear) - : "flags"); + : INT80_CLOBBERS); #ifdef __x86_64__ syscall(SYS_arch_prctl, ARCH_GET_GS, &new_base); diff --git a/tools/testing/selftests/x86/ptrace_syscall.c b/tools/testing/selftests/x86/ptrace_syscall.c index b037ce9cf116..eaea92439708 100644 --- a/tools/testing/selftests/x86/ptrace_syscall.c +++ b/tools/testing/selftests/x86/ptrace_syscall.c @@ -58,7 +58,8 @@ static void do_full_int80(struct syscall_args32 *args) asm volatile ("int $0x80" : "+a" (args->nr), "+b" (args->arg0), "+c" (args->arg1), "+d" (args->arg2), - "+S" (args->arg3), "+D" (args->arg4), "+r" (bp)); + "+S" (args->arg3), "+D" (args->arg4), "+r" (bp) + : : "r8", "r9", "r10", "r11"); args->arg5 = bp; #else sys32_helper(args, int80_and_ret); diff --git a/tools/testing/selftests/x86/single_step_syscall.c b/tools/testing/selftests/x86/single_step_syscall.c index 50c26358e8b7..a48da95c18fd 100644 --- a/tools/testing/selftests/x86/single_step_syscall.c +++ b/tools/testing/selftests/x86/single_step_syscall.c @@ -56,9 +56,11 @@ static volatile sig_atomic_t sig_traps; #ifdef __x86_64__ # define REG_IP REG_RIP # define WIDTH "q" +# define INT80_CLOBBERS "r8", "r9", "r10", "r11" #else # define REG_IP REG_EIP # define WIDTH "l" +# define INT80_CLOBBERS #endif static unsigned long get_eflags(void) @@ -140,7 +142,8 @@ int main() printf("[RUN]\tSet TF and check int80\n"); set_eflags(get_eflags() | X86_EFLAGS_TF); - asm volatile ("int $0x80" : "=a" (tmp) : "a" (SYS_getpid)); + asm volatile ("int $0x80" : "=a" (tmp) : "a" (SYS_getpid) + : INT80_CLOBBERS); check_result(); /* -- 2.11.0 -- To unsubscribe from this list: send the line "unsubscribe linux-kselftest" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html

7 years, 6 months

2
2
0 0

[Linux-kselftest-mirror] selftests: x86_64: kselftest mpx-mini-test_64 - no MPX support - failed - 3869 Aborted (core dumped)

by Naresh Kamboju

Hi, We have found "no MPX support" bug while running kselftests on SuperServer 5019S-ML - x86_64 device. Please provide your comments to fix this bug. LKFT: 4.4-rc 4.9-rc 4.13-rc 4.14-rc: x86: kselftest mpx-mini-test_64 - no MPX support - failed - 3869 Aborted (core dumped) (edit) https://bugs.linaro.org/show_bug.cgi?id=3497 NOTE: Please add your self to Bugzilla to add comments. Best regards Naresh Kamboju -- To unsubscribe from this list: send the line "unsubscribe linux-kselftest" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html

7 years, 6 months

3
6
0 0

[Linux-kselftest-mirror] [PATCH 2/2] [RFC] packet: experimental support for 64-bit timestamps

by Arnd Bergmann

I tried to figure out what it would take to do a version 4 mmap packet socket interface to completely avoid the y2106 overflow problem. This is what I came up with, reusing most of the v3 code, except for the parts where we access the timestamps. For kselftest, I'm adding support for testing v4 in addition to v1-v3, but the test currently does not look at the timestamps, so it won't check that the timestamp format actually works as intended, only that I didn't break the parts that worked in the v3 selftest. Overall, this is more of a mess than I expected, so it's probably not worth doing a v4 format just for the timestamp, but the patch can serve as a reference for anyone that needs a new format for other reasons and fixes this along with the other changes. Signed-off-by: Arnd Bergmann <arnd(a)arndb.de> --- Untested and rather invasive, so don't apply this part without discussion and testing --- include/uapi/linux/if_packet.h | 24 +++++- net/packet/af_packet.c | 115 ++++++++++++++++++++-------- tools/testing/selftests/net/psock_tpacket.c | 65 +++++++++------- 3 files changed, 142 insertions(+), 62 deletions(-) diff --git a/include/uapi/linux/if_packet.h b/include/uapi/linux/if_packet.h index 67b61d91d89b..c2cf29acdd40 100644 --- a/include/uapi/linux/if_packet.h +++ b/include/uapi/linux/if_packet.h @@ -177,8 +177,27 @@ struct tpacket3_hdr { __u8 tp_padding[8]; }; +struct tpacket4_hdr { + __u32 tp_next_offset; + __u32 tp_nsec_hi; + __u32 tp_nsec_lo; + __u32 tp_snaplen; + __u32 tp_len; + __u32 tp_status; + __u16 tp_mac; + __u16 tp_net; + /* pkt_hdr variants */ + union { + struct tpacket_hdr_variant1 hv1; + }; + __u8 tp_padding[8]; +}; + struct tpacket_bd_ts { - unsigned int ts_sec; + union { + unsigned int ts_nsec_hi; + unsigned int ts_sec; + }; union { unsigned int ts_usec; unsigned int ts_nsec; @@ -250,7 +269,8 @@ struct tpacket_block_desc { enum tpacket_versions { TPACKET_V1, TPACKET_V2, - TPACKET_V3 + TPACKET_V3, + TPACKET_V4, }; /* diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c index 7432c6699818..34a07e4a93a5 100644 --- a/net/packet/af_packet.c +++ b/net/packet/af_packet.c @@ -164,6 +164,7 @@ union tpacket_uhdr { struct tpacket_hdr *h1; struct tpacket2_hdr *h2; struct tpacket3_hdr *h3; + struct tpacket4_hdr *h4; void *raw; }; @@ -200,7 +201,7 @@ static void prb_retire_current_block(struct tpacket_kbdq_core *, struct packet_sock *, unsigned int status); static int prb_queue_frozen(struct tpacket_kbdq_core *); static void prb_open_block(struct tpacket_kbdq_core *, - struct tpacket_block_desc *); + struct tpacket_block_desc *, struct packet_sock *po); static void prb_retire_rx_blk_timer_expired(struct timer_list *); static void _prb_refresh_rx_retire_blk_timer(struct tpacket_kbdq_core *); static void prb_fill_rxhash(struct tpacket_kbdq_core *, struct tpacket3_hdr *); @@ -404,6 +405,7 @@ static void __packet_set_status(struct packet_sock *po, void *frame, int status) flush_dcache_page(pgv_to_page(&h.h2->tp_status)); break; case TPACKET_V3: + case TPACKET_V4: h.h3->tp_status = status; flush_dcache_page(pgv_to_page(&h.h3->tp_status)); break; @@ -430,6 +432,7 @@ static int __packet_get_status(struct packet_sock *po, void *frame) flush_dcache_page(pgv_to_page(&h.h2->tp_status)); return h.h2->tp_status; case TPACKET_V3: + case TPACKET_V4: flush_dcache_page(pgv_to_page(&h.h3->tp_status)); return h.h3->tp_status; default: @@ -439,17 +442,17 @@ static int __packet_get_status(struct packet_sock *po, void *frame) } } -static __u32 tpacket_get_timestamp(struct sk_buff *skb, struct timespec64 *ts, +static __u32 tpacket_get_timestamp(struct sk_buff *skb, s64 *stamp, unsigned int flags) { struct skb_shared_hwtstamps *shhwtstamps = skb_hwtstamps(skb); if (shhwtstamps && (flags & SOF_TIMESTAMPING_RAW_HARDWARE) && - ktime_to_timespec64_cond(shhwtstamps->hwtstamp, ts)) + (*stamp = ktime_to_ns(shhwtstamps->hwtstamp))) return TP_STATUS_TS_RAW_HARDWARE; - if (ktime_to_timespec64_cond(skb->tstamp, ts)) + if ((*stamp = ktime_to_ns(skb->tstamp))) return TP_STATUS_TS_SOFTWARE; return 0; @@ -460,19 +463,15 @@ static __u32 __packet_set_timestamp(struct packet_sock *po, void *frame, { union tpacket_uhdr h; struct timespec64 ts; + s64 stamp; __u32 ts_status; - if (!(ts_status = tpacket_get_timestamp(skb, &ts, po->tp_tstamp))) + if (!(ts_status = tpacket_get_timestamp(skb, &stamp, po->tp_tstamp))) return 0; h.raw = frame; - /* - * versions 1 through 3 overflow the timestamps in y2106, since they - * all store the seconds in a 32-bit unsigned integer. - * If we create a version 4, that should have a 64-bit timestamp, - * either 64-bit seconds + 32-bit nanoseconds, or just 64-bit - * nanoseconds. - */ + + ts = ns_to_timespec64(stamp); switch (po->tp_version) { case TPACKET_V1: h.h1->tp_sec = ts.tv_sec; @@ -486,6 +485,9 @@ static __u32 __packet_set_timestamp(struct packet_sock *po, void *frame, h.h3->tp_sec = ts.tv_sec; h.h3->tp_nsec = ts.tv_nsec; break; + case TPACKET_V4: + h.h4->tp_nsec_hi = upper_32_bits(stamp); + h.h4->tp_nsec_lo = lower_32_bits(stamp); default: WARN(1, "TPACKET version not supported.\n"); BUG(); @@ -633,7 +635,7 @@ static void init_prb_bdqc(struct packet_sock *po, p1->max_frame_len = p1->kblk_size - BLK_PLUS_PRIV(p1->blk_sizeof_priv); prb_init_ft_ops(p1, req_u); prb_setup_retire_blk_timer(po); - prb_open_block(p1, pbd); + prb_open_block(p1, pbd, po); } /* Do NOT update the last_blk_num first. @@ -730,7 +732,7 @@ static void prb_retire_rx_blk_timer_expired(struct timer_list *t) * opening a block thaws the queue,restarts timer * Thawing/timer-refresh is a side effect. */ - prb_open_block(pkc, pbd); + prb_open_block(pkc, pbd, po); goto out; } } @@ -792,30 +794,43 @@ static void prb_close_block(struct tpacket_kbdq_core *pkc1, { __u32 status = TP_STATUS_USER | stat; - struct tpacket3_hdr *last_pkt; + struct tpacket3_hdr *last_pkt_v3; + struct tpacket4_hdr *last_pkt_v4; struct tpacket_hdr_v1 *h1 = &pbd1->hdr.bh1; struct sock *sk = &po->sk; if (po->stats.stats3.tp_drops) status |= TP_STATUS_LOSING; - last_pkt = (struct tpacket3_hdr *)pkc1->prev; - last_pkt->tp_next_offset = 0; + last_pkt_v3 = (struct tpacket3_hdr *)pkc1->prev; + last_pkt_v4 = (struct tpacket4_hdr *)pkc1->prev; + last_pkt_v3->tp_next_offset = 0; /* Get the ts of the last pkt */ if (BLOCK_NUM_PKTS(pbd1)) { - h1->ts_last_pkt.ts_sec = last_pkt->tp_sec; - h1->ts_last_pkt.ts_nsec = last_pkt->tp_nsec; + if (po->tp_version == TPACKET_V3) { + h1->ts_last_pkt.ts_sec = last_pkt_v3->tp_sec; + h1->ts_last_pkt.ts_nsec = last_pkt_v3->tp_nsec; + } else { + h1->ts_last_pkt.ts_nsec_hi = last_pkt_v4->tp_nsec_hi; + h1->ts_last_pkt.ts_nsec = last_pkt_v4->tp_nsec_lo; + } } else { /* Ok, we tmo'd - so get the current time. * * It shouldn't really happen as we don't close empty * blocks. See prb_retire_rx_blk_timer_expired(). */ - struct timespec64 ts; - ktime_get_real_ts64(&ts); - h1->ts_last_pkt.ts_sec = ts.tv_sec; - h1->ts_last_pkt.ts_nsec = ts.tv_nsec; + if (po->tp_version == TPACKET_V3) { + struct timespec64 ts; + ktime_get_real_ts64(&ts); + h1->ts_last_pkt.ts_sec = ts.tv_sec; + h1->ts_last_pkt.ts_nsec = ts.tv_nsec; + } else { + u64 ns = ktime_get_real_ns(); + h1->ts_last_pkt.ts_nsec_hi = upper_32_bits(ns); + h1->ts_last_pkt.ts_nsec = lower_32_bits(ns); + } } smp_wmb(); @@ -841,9 +856,8 @@ static void prb_thaw_queue(struct tpacket_kbdq_core *pkc) * */ static void prb_open_block(struct tpacket_kbdq_core *pkc1, - struct tpacket_block_desc *pbd1) + struct tpacket_block_desc *pbd1, struct packet_sock *po) { - struct timespec64 ts; struct tpacket_hdr_v1 *h1 = &pbd1->hdr.bh1; smp_rmb(); @@ -856,10 +870,19 @@ static void prb_open_block(struct tpacket_kbdq_core *pkc1, BLOCK_NUM_PKTS(pbd1) = 0; BLOCK_LEN(pbd1) = BLK_PLUS_PRIV(pkc1->blk_sizeof_priv); - ktime_get_real_ts64(&ts); - h1->ts_first_pkt.ts_sec = ts.tv_sec; - h1->ts_first_pkt.ts_nsec = ts.tv_nsec; + if (po->tp_version == TPACKET_V3) { + struct timespec64 ts; + + ktime_get_real_ts64(&ts); + h1->ts_first_pkt.ts_sec = ts.tv_sec; + h1->ts_first_pkt.ts_nsec = ts.tv_nsec; + } else { + s64 ns = ktime_get_real_ns(); + + h1->ts_first_pkt.ts_nsec_hi = upper_32_bits(ns); + h1->ts_first_pkt.ts_nsec = lower_32_bits(ns); + } pkc1->pkblk_start = (char *)pbd1; pkc1->nxt_offset = pkc1->pkblk_start + BLK_PLUS_PRIV(pkc1->blk_sizeof_priv); @@ -936,7 +959,7 @@ static void *prb_dispatch_next_block(struct tpacket_kbdq_core *pkc, * open this block and return the offset where the first packet * needs to get stored. */ - prb_open_block(pkc, pbd); + prb_open_block(pkc, pbd, po); return (void *)pkc->nxt_offset; } @@ -1068,7 +1091,7 @@ static void *__packet_lookup_frame_in_block(struct packet_sock *po, * opening a block also thaws the queue. * Thawing is a side effect. */ - prb_open_block(pkc, pbd); + prb_open_block(pkc, pbd, po); } } @@ -1113,6 +1136,7 @@ static void *packet_current_rx_frame(struct packet_sock *po, po->rx_ring.head, status); return curr; case TPACKET_V3: + case TPACKET_V4: return __packet_lookup_frame_in_block(po, skb, status, len); default: WARN(1, "TPACKET version not supported\n"); @@ -1171,6 +1195,7 @@ static void packet_increment_rx_head(struct packet_sock *po, case TPACKET_V2: return packet_increment_head(rb); case TPACKET_V3: + case TPACKET_V4: default: WARN(1, "TPACKET version not supported.\n"); BUG(); @@ -1279,7 +1304,7 @@ static int __packet_rcv_has_room(struct packet_sock *po, struct sk_buff *skb) return ROOM_NONE; } - if (po->tp_version == TPACKET_V3) { + if (po->tp_version == TPACKET_V3 || po->tp_version == TPACKET_V4) { if (__tpacket_v3_has_room(po, ROOM_POW_OFF)) ret = ROOM_NORMAL; else if (__tpacket_v3_has_room(po, 0)) @@ -2192,6 +2217,7 @@ static int tpacket_rcv(struct sk_buff *skb, struct net_device *dev, unsigned short macoff, netoff, hdrlen; struct sk_buff *copy_skb = NULL; struct timespec64 ts; + u64 ns; __u32 ts_status; bool is_drop_n_account = false; bool do_vnet = false; @@ -2318,7 +2344,9 @@ static int tpacket_rcv(struct sk_buff *skb, struct net_device *dev, skb_copy_bits(skb, 0, h.raw + macoff, snaplen); - if (!(ts_status = tpacket_get_timestamp(skb, &ts, po->tp_tstamp))) + if ((ts_status = tpacket_get_timestamp(skb, &ns, po->tp_tstamp))) + ts = ns_to_timespec64(ns); + else ktime_get_real_ts64(&ts); status |= ts_status; @@ -2365,6 +2393,19 @@ static int tpacket_rcv(struct sk_buff *skb, struct net_device *dev, memset(h.h3->tp_padding, 0, sizeof(h.h3->tp_padding)); hdrlen = sizeof(*h.h3); break; + case TPACKET_V4: + /* identical to v3, except for the timestamp */ + h.h4->tp_status |= status; + h.h4->tp_len = skb->len; + h.h4->tp_snaplen = snaplen; + h.h4->tp_mac = macoff; + h.h4->tp_net = netoff; + ns = timespec64_to_ns(&ts); + h.h4->tp_nsec_hi = upper_32_bits(ns); + h.h4->tp_nsec_lo = lower_32_bits(ns); + memset(h.h3->tp_padding, 0, sizeof(h.h3->tp_padding)); + hdrlen = sizeof(*h.h3); + break; default: BUG(); } @@ -2570,6 +2611,7 @@ static int tpacket_parse_header(struct packet_sock *po, void *frame, ph.raw = frame; switch (po->tp_version) { + case TPACKET_V4: case TPACKET_V3: if (ph.h3->tp_next_offset != 0) { pr_warn_once("variable sized slot not supported"); @@ -2596,6 +2638,7 @@ static int tpacket_parse_header(struct packet_sock *po, void *frame, off_max = po->tx_ring.frame_size - tp_len; if (po->sk.sk_type == SOCK_DGRAM) { switch (po->tp_version) { + case TPACKET_V4: case TPACKET_V3: off = ph.h3->tp_net; break; @@ -2608,6 +2651,7 @@ static int tpacket_parse_header(struct packet_sock *po, void *frame, } } else { switch (po->tp_version) { + case TPACKET_V4: case TPACKET_V3: off = ph.h3->tp_mac; break; @@ -3658,6 +3702,7 @@ packet_setsockopt(struct socket *sock, int level, int optname, char __user *optv len = sizeof(req_u.req); break; case TPACKET_V3: + case TPACKET_V4: default: len = sizeof(req_u.req3); break; @@ -3693,6 +3738,7 @@ packet_setsockopt(struct socket *sock, int level, int optname, char __user *optv case TPACKET_V1: case TPACKET_V2: case TPACKET_V3: + case TPACKET_V4: break; default: return -EINVAL; @@ -3868,7 +3914,7 @@ static int packet_getsockopt(struct socket *sock, int level, int optname, memset(&po->stats, 0, sizeof(po->stats)); spin_unlock_bh(&sk->sk_receive_queue.lock); - if (po->tp_version == TPACKET_V3) { + if (po->tp_version == TPACKET_V3 || po->tp_version == TPACKET_V4) { lv = sizeof(struct tpacket_stats_v3); st.stats3.tp_packets += st.stats3.tp_drops; data = &st.stats3; @@ -3906,6 +3952,7 @@ static int packet_getsockopt(struct socket *sock, int level, int optname, val = sizeof(struct tpacket2_hdr); break; case TPACKET_V3: + case TPACKET_V4: val = sizeof(struct tpacket3_hdr); break; default: @@ -4250,6 +4297,7 @@ static int packet_set_ring(struct sock *sk, union tpacket_req_u *req_u, po->tp_hdrlen = TPACKET2_HDRLEN; break; case TPACKET_V3: + case TPACKET_V4: po->tp_hdrlen = TPACKET3_HDRLEN; break; } @@ -4284,6 +4332,7 @@ static int packet_set_ring(struct sock *sk, union tpacket_req_u *req_u, if (unlikely(!pg_vec)) goto out; switch (po->tp_version) { + case TPACKET_V4: case TPACKET_V3: /* Block transmit is not supported yet */ if (!tx_ring) { diff --git a/tools/testing/selftests/net/psock_tpacket.c b/tools/testing/selftests/net/psock_tpacket.c index 7f6cd9fdacf3..b497632b6a70 100644 --- a/tools/testing/selftests/net/psock_tpacket.c +++ b/tools/testing/selftests/net/psock_tpacket.c @@ -3,7 +3,8 @@ * Author: Daniel Borkmann <dborkman(a)redhat.com> * Chetan Loke <loke.chetan(a)gmail.com> (TPACKET_V3 usage example) * - * A basic test of packet socket's TPACKET_V1/TPACKET_V2/TPACKET_V3 behavior. + * A basic test of packet socket's TPACKET_V1/TPACKET_V2/TPACKET_V3/TPACKET_V4 + * behavior. * * Control: * Test the setup of the TPACKET socket with different patterns that are @@ -19,6 +20,7 @@ * - TPACKET_V1: RX_RING, TX_RING * - TPACKET_V2: RX_RING, TX_RING * - TPACKET_V3: RX_RING + * - TPACKET_V4: RX_RING * * License (GPLv2): * @@ -310,12 +312,12 @@ static inline void __v2_tx_user_ready(struct tpacket2_hdr *hdr) __sync_synchronize(); } -static inline int __v3_tx_kernel_ready(struct tpacket3_hdr *hdr) +static inline int __v3_v4_tx_kernel_ready(struct tpacket3_hdr *hdr) { return !(hdr->tp_status & (TP_STATUS_SEND_REQUEST | TP_STATUS_SENDING)); } -static inline void __v3_tx_user_ready(struct tpacket3_hdr *hdr) +static inline void __v3_v4_tx_user_ready(struct tpacket3_hdr *hdr) { hdr->tp_status = TP_STATUS_SEND_REQUEST; __sync_synchronize(); @@ -329,7 +331,8 @@ static inline int __tx_kernel_ready(void *base, int version) case TPACKET_V2: return __v2_tx_kernel_ready(base); case TPACKET_V3: - return __v3_tx_kernel_ready(base); + case TPACKET_V4: + return __v3_v4_tx_kernel_ready(base); default: bug_on(1); return 0; @@ -346,7 +349,8 @@ static inline void __tx_user_ready(void *base, int version) __v2_tx_user_ready(base); break; case TPACKET_V3: - __v3_tx_user_ready(base); + case TPACKET_V4: + __v3_v4_tx_user_ready(base); break; } } @@ -372,6 +376,7 @@ static inline void *get_next_frame(struct ring *ring, int n) case TPACKET_V2: return ring->rd[n].iov_base; case TPACKET_V3: + case TPACKET_V4: return f0 + (n * ring->req3.tp_frame_size); default: bug_on(1); @@ -454,7 +459,8 @@ static void walk_tx(int sock, struct ring *ring) packet_len); total_bytes += ppd.v2->tp_h.tp_snaplen; break; - case TPACKET_V3: { + case TPACKET_V3: + case TPACKET_V4: { struct tpacket3_hdr *tx = next; tx->tp_snaplen = packet_len; @@ -517,22 +523,22 @@ static void walk_v1_v2(int sock, struct ring *ring) walk_tx(sock, ring); } -static uint64_t __v3_prev_block_seq_num = 0; +static uint64_t __v3_v4_prev_block_seq_num = 0; -void __v3_test_block_seq_num(struct block_desc *pbd) +void __v3_v4_test_block_seq_num(struct block_desc *pbd) { - if (__v3_prev_block_seq_num + 1 != pbd->h1.seq_num) { + if (__v3_v4_prev_block_seq_num + 1 != pbd->h1.seq_num) { fprintf(stderr, "\nprev_block_seq_num:%"PRIu64", expected " "seq:%"PRIu64" != actual seq:%"PRIu64"\n", - __v3_prev_block_seq_num, __v3_prev_block_seq_num + 1, + __v3_v4_prev_block_seq_num, __v3_v4_prev_block_seq_num + 1, (uint64_t) pbd->h1.seq_num); exit(1); } - __v3_prev_block_seq_num = pbd->h1.seq_num; + __v3_v4_prev_block_seq_num = pbd->h1.seq_num; } -static void __v3_test_block_len(struct block_desc *pbd, uint32_t bytes, int block_num) +static void __v3_v4_test_block_len(struct block_desc *pbd, uint32_t bytes, int block_num) { if (pbd->h1.num_pkts && bytes != pbd->h1.blk_len) { fprintf(stderr, "\nblock:%u with %upackets, expected " @@ -542,23 +548,23 @@ static void __v3_test_block_len(struct block_desc *pbd, uint32_t bytes, int bloc } } -static void __v3_test_block_header(struct block_desc *pbd, const int block_num) +static void __v3_v4_test_block_header(struct block_desc *pbd, const int block_num) { if ((pbd->h1.block_status & TP_STATUS_USER) == 0) { fprintf(stderr, "\nblock %u: not in TP_STATUS_USER\n", block_num); exit(1); } - __v3_test_block_seq_num(pbd); + __v3_v4_test_block_seq_num(pbd); } -static void __v3_walk_block(struct block_desc *pbd, const int block_num) +static void __v3_v4_walk_block(struct block_desc *pbd, const int block_num) { int num_pkts = pbd->h1.num_pkts, i; unsigned long bytes = 0, bytes_with_padding = ALIGN_8(sizeof(*pbd)); struct tpacket3_hdr *ppd; - __v3_test_block_header(pbd, block_num); + __v3_v4_test_block_header(pbd, block_num); ppd = (struct tpacket3_hdr *) ((uint8_t *) pbd + pbd->h1.offset_to_first_pkt); @@ -580,17 +586,17 @@ static void __v3_walk_block(struct block_desc *pbd, const int block_num) __sync_synchronize(); } - __v3_test_block_len(pbd, bytes_with_padding, block_num); + __v3_v4_test_block_len(pbd, bytes_with_padding, block_num); total_bytes += bytes; } -void __v3_flush_block(struct block_desc *pbd) +void __v3_v4_flush_block(struct block_desc *pbd) { pbd->h1.block_status = TP_STATUS_KERNEL; __sync_synchronize(); } -static void walk_v3_rx(int sock, struct ring *ring) +static void walk_v3_v4_rx(int sock, struct ring *ring) { unsigned int block_num = 0; struct pollfd pfd; @@ -614,8 +620,8 @@ static void walk_v3_rx(int sock, struct ring *ring) while ((pbd->h1.block_status & TP_STATUS_USER) == 0) poll(&pfd, 1, 1); - __v3_walk_block(pbd, block_num); - __v3_flush_block(pbd); + __v3_v4_walk_block(pbd, block_num); + __v3_v4_flush_block(pbd); block_num = (block_num + 1) % ring->rd_num; } @@ -623,7 +629,7 @@ static void walk_v3_rx(int sock, struct ring *ring) pair_udp_close(udp_sock); if (total_packets != 2 * NUM_PACKETS) { - fprintf(stderr, "walk_v3_rx: received %u out of %u pkts\n", + fprintf(stderr, "walk_v3_v4_rx: received %u out of %u pkts\n", total_packets, NUM_PACKETS); exit(1); } @@ -631,10 +637,10 @@ static void walk_v3_rx(int sock, struct ring *ring) fprintf(stderr, " %u pkts (%u bytes)", NUM_PACKETS, total_bytes >> 1); } -static void walk_v3(int sock, struct ring *ring) +static void walk_v3_v4(int sock, struct ring *ring) { if (ring->type == PACKET_RX_RING) - walk_v3_rx(sock, ring); + walk_v3_v4_rx(sock, ring); else walk_tx(sock, ring); } @@ -655,7 +661,7 @@ static void __v1_v2_fill(struct ring *ring, unsigned int blocks) ring->flen = ring->req.tp_frame_size; } -static void __v3_fill(struct ring *ring, unsigned int blocks, int type) +static void __v3_v4_fill(struct ring *ring, unsigned int blocks, int type) { if (type == PACKET_RX_RING) { ring->req3.tp_retire_blk_tov = 64; @@ -671,7 +677,7 @@ static void __v3_fill(struct ring *ring, unsigned int blocks, int type) ring->req3.tp_block_nr; ring->mm_len = ring->req3.tp_block_size * ring->req3.tp_block_nr; - ring->walk = walk_v3; + ring->walk = walk_v3_v4; ring->rd_num = ring->req3.tp_block_nr; ring->flen = ring->req3.tp_block_size; } @@ -695,7 +701,8 @@ static void setup_ring(int sock, struct ring *ring, int version, int type) break; case TPACKET_V3: - __v3_fill(ring, blocks, type); + case TPACKET_V4: + __v3_v4_fill(ring, blocks, type); ret = setsockopt(sock, SOL_PACKET, type, &ring->req3, sizeof(ring->req3)); break; @@ -804,6 +811,7 @@ static const char *tpacket_str[] = { [TPACKET_V1] = "TPACKET_V1", [TPACKET_V2] = "TPACKET_V2", [TPACKET_V3] = "TPACKET_V3", + [TPACKET_V4] = "TPACKET_V4", }; static const char *type_str[] = { @@ -854,6 +862,9 @@ int main(void) ret |= test_tpacket(TPACKET_V3, PACKET_RX_RING); ret |= test_tpacket(TPACKET_V3, PACKET_TX_RING); + ret |= test_tpacket(TPACKET_V4, PACKET_RX_RING); + ret |= test_tpacket(TPACKET_V4, PACKET_TX_RING); + if (ret) return 1; -- 2.9.0 -- To unsubscribe from this list: send the line "unsubscribe linux-kselftest" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html

7 years, 7 months

4
7
0 0

[Linux-kselftest-mirror] [PATCH 0/2] selftests: ftrace: Fix multiple-kprobe testcase

by Masami Hiramatsu

Hi, These patches are fixing a bug and improve testcase to ensure adding 256 kprobe events for test. The 1st patch is clearly a bug, so it should be fixed in stable kernel too. Thank you, --- Masami Hiramatsu (2): [BUGFIX] selftest: ftrace: Fix to pick text symbols for kprobes selftest: ftrace: Fix to add 256 kprobe events correctly .../ftrace/test.d/kprobe/multiple_kprobes.tc | 21 +++++++++++++++++--- 1 file changed, 18 insertions(+), 3 deletions(-) -- Masami Hiramatsu (Linaro) <mhiramat(a)kernel.org> -- To unsubscribe from this list: send the line "unsubscribe linux-kselftest" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html

7 years, 7 months

1
2
0 0

Re: [Linux-kselftest-mirror] [RFC PATCH for 4.15 v3 15/22] rseq: selftests: Provide self-tests

by Peter Zijlstra

On Tue, Nov 21, 2017 at 09:18:53AM -0500, Mathieu Desnoyers wrote: > +int percpu_list_push(struct percpu_list *list, struct percpu_list_node *node) > +{ > + intptr_t *targetptr, newval, expect; > + int cpu, ret; > + > + /* Try fast path. */ > + cpu = rseq_cpu_start(); > + /* Load list->c[cpu].head with single-copy atomicity. */ > + expect = (intptr_t)READ_ONCE(list->c[cpu].head); > + newval = (intptr_t)node; > + targetptr = (intptr_t *)&list->c[cpu].head; > + node->next = (struct percpu_list_node *)expect; > + ret = rseq_cmpeqv_storev(targetptr, expect, newval, cpu); > + if (likely(!ret)) > + return cpu; > + return cpu; > +} > +static inline __attribute__((always_inline)) > +int rseq_cmpeqv_storev(intptr_t *v, intptr_t expect, intptr_t newv, > + int cpu) > +{ > + __asm__ __volatile__ goto ( > + RSEQ_ASM_DEFINE_TABLE(3, __rseq_table, 0x0, 0x0, 1f, 2f-1f, 4f) > + RSEQ_ASM_STORE_RSEQ_CS(1, 3b, rseq_cs) > + RSEQ_ASM_CMP_CPU_ID(cpu_id, current_cpu_id, 4f) So the actual C part of the RSEQ is subject to an ABA, right? We can get migrated to another CPU and back again without then failing here. It used to be that this was caught by the sequence count, but that is now gone. The thing that makes it work is the compare against @v: > + "cmpq %[v], %[expect]\n\t" > + "jnz 5f\n\t" That then ensures things are still as we observed them before (although this itself is also subject to ABA). This means all RSEQ primitives that have a C part must have a cmp-and- form, but I suppose that was already pretty much the case anyway. I just don't remember seeing that spelled out anywhere. Then again, I've not yet read that manpage. > + /* final store */ > + "movq %[newv], %[v]\n\t" > + "2:\n\t" > + RSEQ_ASM_DEFINE_ABORT(4, __rseq_failure, RSEQ_SIG, "", abort) > + RSEQ_ASM_DEFINE_CMPFAIL(5, __rseq_failure, "", cmpfail) > + : /* gcc asm goto does not allow outputs */ > + : [cpu_id]"r"(cpu), > + [current_cpu_id]"m"(__rseq_abi.cpu_id), > + [rseq_cs]"m"(__rseq_abi.rseq_cs), > + [v]"m"(*v), > + [expect]"r"(expect), > + [newv]"r"(newv) > + : "memory", "cc", "rax" > + : abort, cmpfail > + ); > + return 0; > +abort: > + return -1; > +cmpfail: > + return 1; > +} -- To unsubscribe from this list: send the line "unsubscribe linux-kselftest" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html

7 years, 7 months

2
3
0 0

Re: [Linux-kselftest-mirror] [RFC PATCH for 4.15 v3 15/22] rseq: selftests: Provide self-tests

by Peter Zijlstra

On Tue, Nov 21, 2017 at 09:18:53AM -0500, Mathieu Desnoyers wrote: > diff --git a/tools/testing/selftests/rseq/rseq-x86.h b/tools/testing/selftests/rseq/rseq-x86.h > new file mode 100644 > index 000000000000..63e81d6c61fa > --- /dev/null > +++ b/tools/testing/selftests/rseq/rseq-x86.h > @@ -0,0 +1,898 @@ > +/* > + * rseq-x86.h > + * > + * (C) Copyright 2016 - Mathieu Desnoyers <mathieu.desnoyers(a)efficios.com> > + * > + * Permission is hereby granted, free of charge, to any person obtaining a copy > + * of this software and associated documentation files (the "Software"), to deal > + * in the Software without restriction, including without limitation the rights > + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell > + * copies of the Software, and to permit persons to whom the Software is > + * furnished to do so, subject to the following conditions: > + * > + * The above copyright notice and this permission notice shall be included in > + * all copies or substantial portions of the Software. > + * > + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR > + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, > + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE > + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER > + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, > + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE > + * SOFTWARE. > + */ > + > +#include <stdint.h> > + > +#define RSEQ_SIG 0x53053053 > + > +#ifdef __x86_64__ > + > +#define rseq_smp_mb() __asm__ __volatile__ ("mfence" : : : "memory") See commit: 450cbdd0125c ("locking/x86: Use LOCK ADD for smp_mb() instead of MFENCE") > +#define rseq_smp_rmb() barrier() > +#define rseq_smp_wmb() barrier() > + > +#define rseq_smp_load_acquire(p) \ > +__extension__ ({ \ > + __typeof(*p) ____p1 = RSEQ_READ_ONCE(*p); \ > + barrier(); \ > + ____p1; \ > +}) > + > +#define rseq_smp_acquire__after_ctrl_dep() rseq_smp_rmb() > + > +#define rseq_smp_store_release(p, v) \ > +do { \ > + barrier(); \ > + RSEQ_WRITE_ONCE(*p, v); \ > +} while (0) > + > +#define RSEQ_ASM_DEFINE_TABLE(label, section, version, flags, \ > + start_ip, post_commit_offset, abort_ip) \ > + ".pushsection " __rseq_str(section) ", \"aw\"\n\t" \ > + ".balign 32\n\t" \ > + __rseq_str(label) ":\n\t" \ > + ".long " __rseq_str(version) ", " __rseq_str(flags) "\n\t" \ > + ".quad " __rseq_str(start_ip) ", " __rseq_str(post_commit_offset) ", " __rseq_str(abort_ip) "\n\t" \ > + ".popsection\n\t" OK, so this creates table entry, but why is @section an argument, AFAICT its _always_ the same thing, no? > +#define RSEQ_ASM_STORE_RSEQ_CS(label, cs_label, rseq_cs) \ > + RSEQ_INJECT_ASM(1) \ > + "leaq " __rseq_str(cs_label) "(%%rip), %%rax\n\t" \ > + "movq %%rax, %[" __rseq_str(rseq_cs) "]\n\t" \ > + __rseq_str(label) ":\n\t" And this sets the TLS variable to point to the table entry from the previous macro, no? But again @rseq_cs seems to always be the very same, why is that an argument? > +#define RSEQ_ASM_CMP_CPU_ID(cpu_id, current_cpu_id, label) \ > + RSEQ_INJECT_ASM(2) \ > + "cmpl %[" __rseq_str(cpu_id) "], %[" __rseq_str(current_cpu_id) "]\n\t" \ > + "jnz " __rseq_str(label) "\n\t" more things that are always the same it seems.. > +#define RSEQ_ASM_DEFINE_ABORT(label, section, sig, teardown, abort_label) \ > + ".pushsection " __rseq_str(section) ", \"ax\"\n\t" \ > + /* Disassembler-friendly signature: nopl <sig>(%rip). */\ > + ".byte 0x0f, 0x1f, 0x05\n\t" \ > + ".long " __rseq_str(sig) "\n\t" \ > + __rseq_str(label) ":\n\t" \ > + teardown \ > + "jmp %l[" __rseq_str(abort_label) "]\n\t" \ > + ".popsection\n\t" @section and @sig seem to always be the same... > +#define RSEQ_ASM_DEFINE_CMPFAIL(label, section, teardown, cmpfail_label) \ > + ".pushsection " __rseq_str(section) ", \"ax\"\n\t" \ > + __rseq_str(label) ":\n\t" \ > + teardown \ > + "jmp %l[" __rseq_str(cmpfail_label) "]\n\t" \ > + ".popsection\n\t" Somewhat failing to see the point of this macro, it seems to just obfuscate the normal failure path. > +static inline __attribute__((always_inline)) > +int rseq_cmpeqv_storev(intptr_t *v, intptr_t expect, intptr_t newv, > + int cpu) I find this a very confusing name for what is essentially compare-and-exchange or compare-and-swap, no? > +{ > + __asm__ __volatile__ goto ( > + RSEQ_ASM_DEFINE_TABLE(3, __rseq_table, 0x0, 0x0, 1f, 2f-1f, 4f) So we set up the section, but unreadably so... reducing the number of arguments would help a lot. Rename the current one to __RSEQ_ASM_DEFINE_TABLE() and then use: #define RSEQ_ASM_DEFINE_TABLE(label, start_ip, post_commit_ip, abort_ip) \ __RSEQ_ASM_DEFINE_TABLE(label, __rseq_table, 0x0, 0x0, start_ip, \ (post_commit_ip - start_ip), abort_ip) or something, such that we can write: RSEQ_ASM_DEFINE_TABLE(3, 1f, 2f, 4f) /* start, commit, abort */ > + RSEQ_ASM_STORE_RSEQ_CS(1, 3b, rseq_cs) And here we open start the rseq by storing the table entry pointer into the TLS thingy. > + RSEQ_ASM_CMP_CPU_ID(cpu_id, current_cpu_id, 4f) > + "cmpq %[v], %[expect]\n\t" > + "jnz 5f\n\t" "jnz %l[cmpfail]\n\t" was too complicated? > + /* final store */ > + "movq %[newv], %[v]\n\t" > + "2:\n\t" > + RSEQ_ASM_DEFINE_ABORT(4, __rseq_failure, RSEQ_SIG, "", abort) > + RSEQ_ASM_DEFINE_CMPFAIL(5, __rseq_failure, "", cmpfail) > + : /* gcc asm goto does not allow outputs */ > + : [cpu_id]"r"(cpu), > + [current_cpu_id]"m"(__rseq_abi.cpu_id), > + [rseq_cs]"m"(__rseq_abi.rseq_cs), > + [v]"m"(*v), > + [expect]"r"(expect), > + [newv]"r"(newv) : [cpu_id] "r" (cpu), [current_cpu_id] "m" (__rseq_abi.cpu_id), [rseq_cs] "m" (__rseq_abi.rseq_cs), [v] "m" (*v), [expect] "r" (expect), [newv] "r" (newv) or something does read much better > + : "memory", "cc", "rax" > + : abort, cmpfail > + ); > + return 0; > +abort: > + return -1; > +cmpfail: > + return 1; > +} > + > +static inline __attribute__((always_inline)) > +int rseq_cmpnev_storeoffp_load(intptr_t *v, intptr_t expectnot, > + off_t voffp, intptr_t *load, int cpu) so this thing does what now? It compares @v to @expectnot, when _not_ matching it will store @voffp into @v and load something..? > +{ > + __asm__ __volatile__ goto ( > + RSEQ_ASM_DEFINE_TABLE(3, __rseq_table, 0x0, 0x0, 1f, 2f-1f, 4f) > + RSEQ_ASM_STORE_RSEQ_CS(1, 3b, rseq_cs) > + RSEQ_ASM_CMP_CPU_ID(cpu_id, current_cpu_id, 4f) > + "cmpq %[v], %[expectnot]\n\t" > + "jz 5f\n\t" So I would prefer "je" in this context, or rather: je %l[cmpfail] > + "movq %[v], %%rax\n\t" loads @v in A But it could already have changed since the previous load from cmp, no? Would it not make sense to put this load before the cmp and use A instead? > + "movq %%rax, %[load]\n\t" stores A in @load > + "addq %[voffp], %%rax\n\t" adds @off to A > + "movq (%%rax), %%rax\n\t" loads (A) in A > + /* final store */ > + "movq %%rax, %[v]\n\t" stores A in @v So the whole thing loads @v into @load, adds and offset, dereferences and adds that back in @v, provided @v doesn't match @expected.. whee. > + "2:\n\t" > + RSEQ_ASM_DEFINE_ABORT(4, __rseq_failure, RSEQ_SIG, "", abort) > + RSEQ_ASM_DEFINE_CMPFAIL(5, __rseq_failure, "", cmpfail) > + : /* gcc asm goto does not allow outputs */ > + : [cpu_id]"r"(cpu), > + [current_cpu_id]"m"(__rseq_abi.cpu_id), > + [rseq_cs]"m"(__rseq_abi.rseq_cs), > + /* final store input */ > + [v]"m"(*v), > + [expectnot]"r"(expectnot), > + [voffp]"er"(voffp), > + [load]"m"(*load) > + : "memory", "cc", "rax" > + : abort, cmpfail > + ); > + return 0; > +abort: > + return -1; > +cmpfail: > + return 1; > +} > +#elif __i386__ > + > +/* > + * Support older 32-bit architectures that do not implement fence > + * instructions. > + */ > +#define rseq_smp_mb() \ > + __asm__ __volatile__ ("lock; addl $0,0(%%esp)" : : : "memory") > +#define rseq_smp_rmb() \ > + __asm__ __volatile__ ("lock; addl $0,0(%%esp)" : : : "memory") > +#define rseq_smp_wmb() \ > + __asm__ __volatile__ ("lock; addl $0,0(%%esp)" : : : "memory") Oh shiny, you're supporting that OOSTORE and PPRO_FENCE nonsense? Going by commit: 09df7c4c8097 ("x86: Remove CONFIG_X86_OOSTORE") That smp_wmb() one was an 'optimization' (forced store buffer flush) but not a correctness thing. And we dropped that stuff from the kernel a _long_ time ago. Ideally we'd kill that PPRO_FENCE crap too. -- To unsubscribe from this list: send the line "unsubscribe linux-kselftest" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html

7 years, 7 months

2
1
0 0

Re: [Linux-kselftest-mirror] [RFC PATCH for 4.15 v3 15/22] rseq: selftests: Provide self-tests

by Peter Zijlstra

On Tue, Nov 21, 2017 at 09:18:53AM -0500, Mathieu Desnoyers wrote: > Implements two basic tests of RSEQ functionality, and one more > exhaustive parameterizable test. > > The first, "basic_test" only asserts that RSEQ works moderately > correctly. E.g. that the CPUID pointer works. > > "basic_percpu_ops_test" is a slightly more "realistic" variant, > implementing a few simple per-cpu operations and testing their > correctness. > > "param_test" is a parametrizable restartable sequences test. See > the "--help" output for usage. > > A run_param_test.sh script runs many variants of the parametrizable > tests. > > As part of those tests, a helper library "rseq" implements a user-space > API around restartable sequences. It uses the cpu_opv system call as > fallback when single-stepped by a debugger. It exposes the instruction > pointer addresses where the rseq assembly blocks begin and end, as well > as the associated abort instruction pointer, in the __rseq_table > section. This section allows debuggers may know where to place > breakpoints when single-stepping through assembly blocks which may be > aborted at any point by the kernel. Could I ask you to split this in smaller bits? I'd start with just the rseq library, using only the rseq interface. Then add the whole cpu_opv fallback stuff. Then add the selftests using librseq. As is this is a tad much to read in a single go. -- To unsubscribe from this list: send the line "unsubscribe linux-kselftest" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html

7 years, 7 months

2
1
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-kselftest-mirror