qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: "Philippe Mathieu-Daudé" <philmd@redhat.com>
To: "Alex Bennée" <alex.bennee@linaro.org>
Cc: Peter Maydell <peter.maydell@linaro.org>,
	qemu-arm@nongnu.org,
	Richard Henderson <richard.henderson@linaro.org>,
	qemu-devel@nongnu.org
Subject: Re: [PATCH v1 20/28] tests/tcg/aarch64: userspace system register test
Date: Tue, 17 Mar 2020 12:43:01 +0100	[thread overview]
Message-ID: <17987391-ad3c-5671-831f-0d36aa2c93b2@redhat.com> (raw)
In-Reply-To: <87lfnzjlax.fsf@linaro.org>

On 3/17/20 12:03 PM, Alex Bennée wrote:
> 
> Philippe Mathieu-Daudé <philmd@redhat.com> writes:
> 
>> On 3/16/20 6:21 PM, Alex Bennée wrote:
>>> This tests a bunch of registers that the kernel allows userspace to
>>> read including the CPUID registers. We need a SVE aware compiler as we
>>> are testing the id_aa64zfr0_el1 register in the set.
>>> Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
>>> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
>>> Message-Id: <20190205190224.2198-7-alex.bennee@linaro.org>
>>> ---
>>> vgdbstub
>>>     - don't build unless using docker or CROSS_CC_HAS_SVE
>>> ---
>>>    tests/tcg/aarch64/sysregs.c       | 172 ++++++++++++++++++++++++++++++
>>>    tests/tcg/aarch64/Makefile.target |   6 ++
>>>    2 files changed, 178 insertions(+)
>>>    create mode 100644 tests/tcg/aarch64/sysregs.c
>>> diff --git a/tests/tcg/aarch64/sysregs.c
>>> b/tests/tcg/aarch64/sysregs.c
>>> new file mode 100644
>>> index 00000000000..40cf8d2877e
>>> --- /dev/null
>>> +++ b/tests/tcg/aarch64/sysregs.c
>>> @@ -0,0 +1,172 @@
>>> +/*
>>> + * Check emulated system register access for linux-user mode.
>>> + *
>>> + * See: https://www.kernel.org/doc/Documentation/arm64/cpu-feature-registers.txt
>>> + *
>>> + * Copyright (c) 2019 Linaro
>>> + *
>>> + * This work is licensed under the terms of the GNU GPL, version 2 or later.
>>> + * See the COPYING file in the top-level directory.
>>> + *
>>> + * SPDX-License-Identifier: GPL-2.0-or-later
>>> + */
>>> +
>>> +#include <asm/hwcap.h>
>>> +#include <stdio.h>
>>> +#include <sys/auxv.h>
>>> +#include <signal.h>
>>> +#include <string.h>
>>> +#include <stdbool.h>
>>> +
>>> +#ifndef HWCAP_CPUID
>>> +#define HWCAP_CPUID (1 << 11)
>>> +#endif
>>> +
>>> +int failed_bit_count;
>>> +
>>> +/* Read and print system register `id' value */
>>> +#define get_cpu_reg(id) ({                                      \
>>> +            unsigned long __val = 0xdeadbeef;                   \
>>> +            asm("mrs %0, "#id : "=r" (__val));                  \
>>> +            printf("%-20s: 0x%016lx\n", #id, __val);            \
>>> +            __val;                                               \
>>> +        })
>>> +
>>> +/* As above but also check no bits outside of `mask' are set*/
>>> +#define get_cpu_reg_check_mask(id, mask) ({                     \
>>> +            unsigned long __cval = get_cpu_reg(id);             \
>>> +            unsigned long __extra = __cval & ~mask;             \
>>> +            if (__extra) {                                      \
>>> +                printf("%-20s: 0x%016lx\n", "  !!extra bits!!", __extra);   \
>>> +                failed_bit_count++;                            \
>>> +            }                                                   \
>>> +})
>>> +
>>> +/* As above but check RAZ */
>>> +#define get_cpu_reg_check_zero(id) ({                           \
>>> +            unsigned long __val = 0xdeadbeef;                   \
>>> +            asm("mrs %0, "#id : "=r" (__val));                  \
>>> +            if (__val) {                                        \
>>> +                printf("%-20s: 0x%016lx (not RAZ!)\n", #id, __val);        \
>>> +                failed_bit_count++;                            \
>>> +            }                                                   \
>>> +})
>>> +
>>> +/* Chunk up mask into 63:48, 47:32, 31:16, 15:0 to ease counting */
>>> +#define _m(a, b, c, d) (0x ## a ## b ## c ## d ##ULL)
>>> +
>>> +bool should_fail;
>>> +int should_fail_count;
>>> +int should_not_fail_count;
>>> +uintptr_t failed_pc[10];
>>> +
>>> +void sigill_handler(int signo, siginfo_t *si, void *data)
>>> +{
>>> +    ucontext_t *uc = (ucontext_t *)data;
>>> +
>>> +    if (should_fail) {
>>> +        should_fail_count++;
>>> +    } else {
>>> +        uintptr_t pc = (uintptr_t) uc->uc_mcontext.pc;
>>> +        failed_pc[should_not_fail_count++] =  pc;
>>> +    }
>>> +    uc->uc_mcontext.pc += 4;
>>> +}
>>> +
>>> +int main(void)
>>> +{
>>> +    struct sigaction sa;
>>> +
>>> +    /* Hook in a SIGILL handler */
>>> +    memset(&sa, 0, sizeof(struct sigaction));
>>> +    sa.sa_flags = SA_SIGINFO;
>>> +    sa.sa_sigaction = &sigill_handler;
>>> +    sigemptyset(&sa.sa_mask);
>>> +
>>> +    if (sigaction(SIGILL, &sa, 0) != 0) {
>>> +        perror("sigaction");
>>> +        return 1;
>>> +    }
>>> +
>>> +    /* Counter values have been exposed since Linux 4.12 */
>>> +    printf("Checking Counter registers\n");
>>> +
>>> +    get_cpu_reg(ctr_el0);
>>> +    get_cpu_reg(cntvct_el0);
>>> +    get_cpu_reg(cntfrq_el0);
>>> +
>>> +    /* HWCAP_CPUID indicates we can read feature registers, since Linux 4.11 */
>>> +    if (!(getauxval(AT_HWCAP) & HWCAP_CPUID)) {
>>> +        printf("CPUID registers unavailable\n");
>>> +        return 1;
>>> +    } else {
>>> +        printf("Checking CPUID registers\n");
>>> +    }
>>> +
>>> +    /*
>>> +     * Some registers only expose some bits to user-space. Anything
>>> +     * that is IMPDEF is exported as 0 to user-space. The _mask checks
>>> +     * assert no extra bits are set.
>>> +     *
>>> +     * This check is *not* comprehensive as some fields are set to
>>> +     * minimum valid fields - for the purposes of this check allowed
>>> +     * to have non-zero values.
>>> +     */
>>> +    get_cpu_reg_check_mask(id_aa64isar0_el1, _m(00ff,ffff,f0ff,fff0));
>>> +    get_cpu_reg_check_mask(id_aa64isar1_el1, _m(0000,00f0,ffff,ffff));
>>> +    /* TGran4 & TGran64 as pegged to -1 */
>>> +    get_cpu_reg_check_mask(id_aa64mmfr0_el1, _m(0000,0000,ff00,0000));
>>> +    get_cpu_reg_check_zero(id_aa64mmfr1_el1);
>>> +    /* EL1/EL0 reported as AA64 only */
>>> +    get_cpu_reg_check_mask(id_aa64pfr0_el1,  _m(000f,000f,00ff,0011));
>>> +    get_cpu_reg_check_mask(id_aa64pfr1_el1,  _m(0000,0000,0000,00f0));
>>> +    /* all hidden, DebugVer fixed to 0x6 (ARMv8 debug architecture) */
>>> +    get_cpu_reg_check_mask(id_aa64dfr0_el1,  _m(0000,0000,0000,0006));
>>> +    get_cpu_reg_check_zero(id_aa64dfr1_el1);
>>> +    get_cpu_reg_check_zero(id_aa64zfr0_el1);
>>> +
>>> +    get_cpu_reg_check_zero(id_aa64afr0_el1);
>>> +    get_cpu_reg_check_zero(id_aa64afr1_el1);
>>> +
>>> +    get_cpu_reg_check_mask(midr_el1,         _m(0000,0000,ffff,ffff));
>>> +    /* mpidr sets bit 31, everything else hidden */
>>> +    get_cpu_reg_check_mask(mpidr_el1,        _m(0000,0000,8000,0000));
>>> +    /* REVIDR is all IMPDEF so should be all zeros to user-space */
>>> +    get_cpu_reg_check_zero(revidr_el1);
>>> +
>>> +    /*
>>> +     * There are a block of more registers that are RAZ in the rest of
>>> +     * the Op0=3, Op1=0, CRn=0, CRm=0,4,5,6,7 space. However for
>>> +     * brevity we don't check stuff that is currently un-allocated
>>> +     * here. Feel free to add them ;-)
>>> +     */
>>> +
>>> +    printf("Remaining registers should fail\n");
>>> +    should_fail = true;
>>> +
>>> +    /* Unexposed register access causes SIGILL */
>>> +    get_cpu_reg(id_mmfr0_el1);
>>> +    get_cpu_reg(id_mmfr1_el1);
>>> +    get_cpu_reg(id_mmfr2_el1);
>>> +    get_cpu_reg(id_mmfr3_el1);
>>> +
>>> +    get_cpu_reg(mvfr0_el1);
>>> +    get_cpu_reg(mvfr1_el1);
>>> +
>>> +    if (should_not_fail_count > 0) {
>>> +        int i;
>>> +        for (i = 0; i < should_not_fail_count; i++) {
>>> +            uintptr_t pc = failed_pc[i];
>>> +            uint32_t insn = *(uint32_t *) pc;
>>> +            printf("insn %#x @ %#lx unexpected FAIL\n", insn, pc);
>>> +        }
>>> +        return 1;
>>> +    }
>>> +
>>> +    if (failed_bit_count > 0) {
>>> +        printf("Extra information leaked to user-space!\n");
>>> +        return 1;
>>> +    }
>>> +
>>> +    return should_fail_count == 6 ? 0 : 1;
>>> +}
>>> diff --git a/tests/tcg/aarch64/Makefile.target b/tests/tcg/aarch64/Makefile.target
>>> index 8ed477d0d51..a25afc071cc 100644
>>> --- a/tests/tcg/aarch64/Makefile.target
>>> +++ b/tests/tcg/aarch64/Makefile.target
>>> @@ -42,4 +42,10 @@ run-semiconsole: semiconsole
>>>    run-plugin-semiconsole-with-%:
>>>    	$(call skip-test, $<, "MANUAL ONLY")
>>>    +ifneq ($(DOCKER_IMAGE)$(CROSS_CC_HAS_SVE),)
>>> +# System Registers Tests
>>> +AARCH64_TESTS += sysregs
>>> +sysregs: CFLAGS+=-march=armv8.1-a+sve
>>> +endif
>>> +
>>>    TESTS += $(AARCH64_TESTS)
>>>
>>
>> warning: while parsing target description (at line 47): Vector
>> "vq128hf" references undefined type "ieee_half"
>> warning: Could not load XML target description; ignoring
>> *** stack smashing detected ***: <unknown> terminated
> 
> How? That test doesn't even use the gdbstub.

This is a bug in my gdb :/

GNU gdb (GDB) Fedora 8.3-7.fc30
Copyright (C) 2019 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later 
<http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.
Type "show copying" and "show warranty" for details.
This GDB was configured as "x86_64-redhat-linux-gnu".
Type "show configuration" for configuration details.
For bug reporting instructions, please see:
<http://www.gnu.org/software/gdb/bugs/>.
Find the GDB manual and other documentation resources online at:
     <http://www.gnu.org/software/gdb/documentation/>.

For help, type "help".
Type "apropos word" to search for commands related to "word"...
Reading symbols from sysregs...
Remote debugging using localhost:1234
warning: while parsing target description (at line 47): Vector "vq128hf" 
references undefined type "ieee_half"
warning: Could not load XML target description; ignoring
0x0000000000400498 in _start ()
PASS: connected to aarch64
x0             0x0                 0
x1             0x0                 0
x2             0x0                 0
x3             0x0                 0
x4             0x0                 0
x5             0x0                 0
x6             0x0                 0
x7             0x0                 0
x8             0x0                 0
x9             0x0                 0
x10            0x0                 0
x11            0x0                 0
x12            0x0                 0
x13            0x0                 0
x14            0x0                 0
x15            0x0                 0
x16            0x0                 0
x17            0x0                 0
x18            0x0                 0
x19            0x0                 0
x20            0x0                 0
x21            0x0                 0
x22            0x0                 0
x23            0x0                 0
x24            0x0                 0
x25            0x0                 0
x26            0x0                 0
x27            0x0                 0
x28            0x0                 0
x29            0x0                 0
x30            0x0                 0
sp             0x40007ff0d0        0x40007ff0d0
pc             0x400498            0x400498 <_start>
cpsr           0x40000000          [ EL=0 Z ]
fpsr           0x0                 0
fpcr           0x0                 0
PASS: info registers
*** stack smashing detected ***: <unknown> terminated
Aborted (core dumped)

(gdb) bt
#0  0x00007f9036e02e35 in raise () from /lib64/libc.so.6
#1  0x00007f9036ded895 in abort () from /lib64/libc.so.6
#2  0x00007f9036e4608f in __libc_message () from /lib64/libc.so.6
#3  0x00007f9036ed6ce5 in __fortify_fail_abort () from /lib64/libc.so.6
#4  0x00007f9036ed6c98 in __stack_chk_fail () from /lib64/libc.so.6
#5  0x0000562619e27191 in remote_target::fetch_register_using_p 
(this=<optimized out>, regcache=0x56261ab84430, reg=0x56261ab09140) at 
../../gdb/remote.c:7959
#6  0x0000000000000000 in ?? ()

> 
>>
>> (GCC 9.2.1, Fedora 30)
> 
> 



  reply	other threads:[~2020-03-17 11:45 UTC|newest]

Thread overview: 46+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-03-16 17:21 [PATCH v1 00/28 for 5.0] testing and gdbstub Alex Bennée
2020-03-16 17:21 ` [PATCH v1 01/28] tests/docker: Install tools to cross-debug and build Linux kernels Alex Bennée
2020-03-16 17:21 ` [PATCH v1 02/28] tests/docker: Update VirGL git repository URL Alex Bennée
2020-03-16 17:21 ` [PATCH v1 03/28] tests/docker: Remove obsolete VirGL --with-glx configure option Alex Bennée
2020-03-16 17:21 ` [PATCH v1 04/28] tests/docker: Update VirGL to v0.8.0 Alex Bennée
2020-03-16 17:21 ` [PATCH v1 05/28] travis.yml: Set G_MESSAGES_DEBUG do report GLib errors Alex Bennée
2020-03-16 17:21 ` [PATCH v1 06/28] gdbstub: make GDBState static and have common init function Alex Bennée
2020-03-16 17:21 ` [PATCH v1 07/28] gdbstub: stop passing GDBState * around and use global Alex Bennée
2020-03-16 17:21 ` [PATCH v1 08/28] gdbstub: move str_buf to GDBState and use GString Alex Bennée
2020-03-16 17:21 ` [PATCH v1 09/28] gdbstub: move mem_buf to GDBState and use GByteArray Alex Bennée
2020-03-16 17:21 ` [PATCH v1 10/28] gdbstub: add helper for 128 bit registers Alex Bennée
2020-03-16 17:21 ` [PATCH v1 11/28] target/arm: use gdb_get_reg helpers Alex Bennée
2020-03-16 17:21 ` [PATCH v1 12/28] target/m68k: " Alex Bennée
2020-03-17  9:48   ` Philippe Mathieu-Daudé
2020-03-16 17:21 ` [PATCH v1 13/28] target/i386: " Alex Bennée
2020-03-17  9:53   ` Philippe Mathieu-Daudé
2020-03-17 10:05     ` Philippe Mathieu-Daudé
2020-03-16 17:21 ` [PATCH v1 14/28] gdbstub: extend GByteArray to read register helpers Alex Bennée
2020-03-16 17:21 ` [PATCH v1 15/28] target/arm: prepare for multiple dynamic XMLs Alex Bennée
2020-03-16 17:21 ` [PATCH v1 16/28] target/arm: explicitly encode regnum in our XML Alex Bennée
2020-03-16 17:21 ` [PATCH v1 17/28] target/arm: default SVE length to 64 bytes for linux-user Alex Bennée
2020-03-16 17:21 ` [PATCH v1 18/28] target/arm: generate xml description of our SVE registers Alex Bennée
2020-03-16 17:21 ` [PATCH v1 19/28] target/arm: don't bother with id_aa64pfr0_read for USER_ONLY Alex Bennée
2020-03-17  9:56   ` Philippe Mathieu-Daudé
2020-03-16 17:21 ` [PATCH v1 20/28] tests/tcg/aarch64: userspace system register test Alex Bennée
2020-03-17 10:24   ` Philippe Mathieu-Daudé
2020-03-17 11:03     ` Alex Bennée
2020-03-17 11:43       ` Philippe Mathieu-Daudé [this message]
2020-03-16 17:21 ` [PATCH v1 21/28] configure: allow user to specify what gdb to use Alex Bennée
2020-03-16 18:27   ` Peter Maydell
2020-03-17  9:46     ` Alex Bennée
2020-03-16 17:21 ` [PATCH v1 22/28] tests/guest-debug: add a simple test runner Alex Bennée
2020-03-17 10:30   ` Philippe Mathieu-Daudé
2020-03-16 17:21 ` [PATCH v1 23/28] tests/tcg/aarch64: add a gdbstub testcase for SVE registers Alex Bennée
2020-03-17 10:30   ` Philippe Mathieu-Daudé
2020-03-16 17:21 ` [PATCH v1 24/28] tests/tcg/aarch64: add SVE iotcl test Alex Bennée
2020-03-17 10:29   ` Philippe Mathieu-Daudé
2020-03-17 10:45     ` Aleksandar Markovic
2020-03-17 10:49       ` Peter Maydell
2020-03-17 13:57         ` Alex Bennée
2020-03-17 13:40     ` Aleksandar Markovic
2020-03-16 17:21 ` [PATCH v1 25/28] tests/tcg/aarch64: add test-sve-ioctl guest-debug test Alex Bennée
2020-03-16 17:21 ` [PATCH v1 26/28] gdbstub: change GDBState.last_packet to GByteArray Alex Bennée
2020-03-16 17:21 ` [PATCH v1 27/28] gdbstub: do not split gdb_monitor_write payload Alex Bennée
2020-03-16 17:21 ` [PATCH v1 28/28] gdbstub: Fix single-step issue by confirming 'vContSupported+' feature to gdb Alex Bennée
2020-03-16 21:26 ` [PATCH v1 00/28 for 5.0] testing and gdbstub no-reply

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=17987391-ad3c-5671-831f-0d36aa2c93b2@redhat.com \
    --to=philmd@redhat.com \
    --cc=alex.bennee@linaro.org \
    --cc=peter.maydell@linaro.org \
    --cc=qemu-arm@nongnu.org \
    --cc=qemu-devel@nongnu.org \
    --cc=richard.henderson@linaro.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).