arm64/sme: ptrace support for TPIDR2_EL0
Subsystems (4)
v4 cover arm64/sme: ptrace support for TPIDR2_EL0
When SME was initially merged we did not add support for TPIDR2_EL0 to the ptrace interface, creating difficulties for debuggers in accessing lazy save state for ZA. This series implements that support, extending the existing NT_ARM_TLS regset to support the register when available, and adds kselftest coverage for the existing and new NT_ARM_TLS functionality.
Existing programs that query the size of the register set will be able to observe the increased size of the register set. Programs that assume the register set is single register will see no change. On systems that do not support SME TPIDR2_EL0 will read as 0 and writes will be ignored, support for SME should be queried via hwcaps as normal.
v4:
- Rebase onto v6.0-rc3.
v3:
- Fix copyright date on test program.
v2:
- Rebase onto v6.0-rc1.
Mark Brown (4):
kselftest/arm64: Add test coverage for NT_ARM_TLS arm64/ptrace: Document extension of NT_ARM_TLS to cover TPIDR2_EL0 arm64/ptrace: Support access to TPIDR2_EL0 kselftest/arm64: Add coverage of TPIDR2_EL0 ptrace interface
Documentation/arm64/sme.rst | 3 + arch/arm64/kernel/ptrace.c | 25 +- tools/testing/selftests/arm64/abi/.gitignore | 1 + tools/testing/selftests/arm64/abi/Makefile | 2 +- tools/testing/selftests/arm64/abi/ptrace.c | 241 +++++++++++++++++++ 5 files changed, 266 insertions(+), 6 deletions(-) create mode 100644 tools/testing/selftests/arm64/abi/ptrace.c
base-commit: b90cb1053190353cc30f0fef0ef1f378ccc063c5
Signature · 1 line
2.30.2
PATCH v4 1/4 kselftest/arm64: Add test coverage for NT_ARM_TLS
In preparation for extending support for NT_ARM_TLS to cover additional TPIDRs add some tests for the existing interface. Do this in a generic ptrace test program to provide a place to collect additional tests in the future.
--- tools/testing/selftests/arm64/abi/.gitignore | 1 + tools/testing/selftests/arm64/abi/Makefile | 2 +- tools/testing/selftests/arm64/abi/ptrace.c | 165 +++++++++++++++++++ 3 files changed, 167 insertions(+), 1 deletion(-) create mode 100644 tools/testing/selftests/arm64/abi/ptrace.c
tools/testing/selftests/arm64/abi/.gitignore +1 -0
@@ -1,2 +1,3 @@ +ptrace syscall-abi tpidr2
tools/testing/selftests/arm64/abi/Makefile +1 -1
@@ -1,7 +1,7 @@ # SPDX-License-Identifier: GPL-2.0 # Copyright (C) 2021 ARM Limited -TEST_GEN_PROGS := syscall-abi tpidr2 +TEST_GEN_PROGS := ptrace syscall-abi tpidr2 include ../../lib.mk
tools/testing/selftests/arm64/abi/ptrace.c new file +165 -0
165 lines not on this page · raw message
Signature · 1 line
2.30.2
PATCH v4 2/4 arm64/ptrace: Document extension of NT_ARM_TLS to cover TPIDR2_EL0
In reply to Mark BrownIn order to allow debuggers to discover lazily saved SME state we need to provide access to TPIDR2_EL0, we will extend the existing NT_ARM_TLS used for TPIDR to also include TPIDR2_EL0 as the second register in the regset.
--- Documentation/arm64/sme.rst | 3 +++ 1 file changed, 3 insertions(+)
Documentation/arm64/sme.rst +3 -0
@@ -331,6 +331,9 @@ The regset data starts with struct user_za_header, containing: been read if a PTRACE_GETREGSET of NT_ARM_ZA were executed for each thread when the coredump was generated. +* The NT_ARM_TLS note will be extended to two registers, the second register + will contain TPIDR2_EL0 on systems that support SME and will be read as + zero with writes ignored otherwise. 9. System runtime configuration --------------------------------
Signature · 1 line
2.30.2
PATCH v4 3/4 arm64/ptrace: Support access to TPIDR2_EL0
In reply to Mark BrownSME introduces an additional EL0 register, TPIDR2_EL0, intended for use by userspace as part of the SME. Provide ptrace access to it through the existing NT_ARM_TLS regset used for TPIDR_EL0 by expanding it to two registers with TPIDR2_EL0 being the second one.
Existing programs that query the size of the register set will be able to observe the increased size of the register set. Programs that assume the register set is single register will see no change. On systems that do not support SME TPIDR2_EL0 will read as 0 and writes will be ignored, support for SME should be queried via hwcaps as normal.
--- arch/arm64/kernel/ptrace.c | 25 ++++++++++++++++++++----- 1 file changed, 20 insertions(+), 5 deletions(-)
arch/arm64/kernel/ptrace.c +20 -5
@@ -666,10 +666,18 @@ static int fpr_set(struct task_struct *target, const struct user_regset *regset,static int tls_get (struct task_struct *target,const struct user_regset *regset,struct membuf to) { +int ret; +if (target == current)tls_preserve_current_state (); -return membuf_store (&to, target->thread.uw.tp_value); + ret =membuf_store (&to, target->thread.uw.tp_value); +if (system_supports_tpidr2 ()) + ret =membuf_store (&to, target->thread.tpidr2_el0); +else + ret =membuf_zero (&to,sizeof (u64)); + +return ret; } static int tls_set(struct task_struct *target, const struct user_regset *regset,
@@ -677,13 +685,20 @@ static int tls_set(struct task_struct *target, const struct user_regset *regset, const void *kbuf, const void __user *ubuf) { int ret; -unsigned long tls = target->thread.uw.tp_value; +unsigned long tls[2 ]; - ret =user_regset_copyin (&pos, &count, &kbuf, &ubuf, &tls,0 ,-1 ); + tls[0 ] = target->thread.uw.tp_value; +if (system_supports_sme ()) + tls[1 ] = target->thread.tpidr2_el0; + + ret =user_regset_copyin (&pos, &count, &kbuf, &ubuf, tls,0 , count);if (ret)return ret; - target->thread.uw.tp_value = tls; + target->thread.uw.tp_value = tls[0 ]; +if (system_supports_sme ()) + target->thread.tpidr2_el0 = tls[1 ]; +return ret; }
@@ -1392,7 +1407,7 @@ static const struct user_regset aarch64_regsets[] = { }, [REGSET_TLS] = { .core_note_type = NT_ARM_TLS, - .n = 1, + .n = 2, .size = sizeof(void *), .align = sizeof(void *), .regset_get = tls_get,
Signature · 1 line
2.30.2
Catalin Marinas
maintainerRe: [PATCH v4 3/4] arm64/ptrace: Support access to TPIDR2_EL0
arch/arm64/kernel/ptrace.c static const struct user_regset aarch64_regsets[] = { · line 1413
Quoted 9 lines · Mark Brown
arch/arm64/kernel/ptrace.c +1
-1@@ -1392,7 +1407,7 @@ static const struct user_regset aarch64_regsets[] = { }, [REGSET_TLS] = { .core_note_type = NT_ARM_TLS, - .n = 1, + .n = 2, .size = sizeof(void *), .align = sizeof(void *), .regset_get = tls_get,
Does this change confuse user-space? I presume an updated gdb would check the iov.len to figure out whether a new register is available but would existing debuggers complain of the new size of this regset?
Signature · 1 line
Catalin
Re: [PATCH v4 3/4] arm64/ptrace: Support access to TPIDR2_EL0
Quoted 14 lines · Catalin Marinas
arch/arm64/kernel/ptrace.c static const struct user_regset aarch64_regsets[] = { · line 1413
Quoted 9 lines · Mark Brown
arch/arm64/kernel/ptrace.c +1
-1@@ -1392,7 +1407,7 @@ static const struct user_regset aarch64_regsets[] = { }, [REGSET_TLS] = { .core_note_type = NT_ARM_TLS, - .n = 1, + .n = 2, .size = sizeof(void *), .align = sizeof(void *), .regset_get = tls_get,Does this change confuse user-space? I presume an updated gdb would check the iov.len to figure out whether a new register is available but would existing debuggers complain of the new size of this regset?
gdb seems happy as far as I can see, it is possible something would be reusing the read_iov for repeated TLS read calls in a context where it was only pointing at a single u64 but I'm not sure how realistic that is given the idiom. I did do a search on sources.debian.net and didn't turn up anything that'd have problems.
If using this as an extensiblility mechanism is a concern we need to bear that in mind elsewhere, and for this it's either a case of providing another single register regset or trying to do a generic sysreg read/get (though that'd be another regset that's not idiomatic for the regset API).
Re: [PATCH v4 3/4] arm64/ptrace: Support access to TPIDR2_EL0
Quoted 22 lines · Mark Brown
Quoted 14 lines · Catalin Marinas
arch/arm64/kernel/ptrace.c static const struct user_regset aarch64_regsets[] = { · line 1413
Quoted 9 lines · Mark Brown
arch/arm64/kernel/ptrace.c +1
-1@@ -1392,7 +1407,7 @@ static const struct user_regset aarch64_regsets[] = { }, [REGSET_TLS] = { .core_note_type = NT_ARM_TLS, - .n = 1, + .n = 2, .size = sizeof(void *), .align = sizeof(void *), .regset_get = tls_get,Does this change confuse user-space? I presume an updated gdb would check the iov.len to figure out whether a new register is available but would existing debuggers complain of the new size of this regset?
gdb seems happy as far as I can see, it is possible something would be reusing the
read_iovfor repeated TLS read calls in a context where it was only pointing at a single u64 but I'm not sure how realistic that is given the idiom. I did do a search on sources.debian.net and didn't turn up anything that'd have problems.
Quoted 5 lines
If using this as an extensiblility mechanism is a concern we need to bear that in mind elsewhere, and for this it's either a case of providing another single register regset or trying to do a generic sysreg read/get (though that'd be another regset that's not idiomatic for the regset API).
Older GDB's assume a single register for NT_ARM_TLS, so they will always fetch TPIDR. Newer GDB's will check the size and act accordingly.
Catalin Marinas
maintainerRe: [PATCH v4 3/4] arm64/ptrace: Support access to TPIDR2_EL0
Quoted 23 lines · Luis Machado
Quoted 22 lines · Mark Brown
Quoted 14 lines · Catalin Marinas
arch/arm64/kernel/ptrace.c static const struct user_regset aarch64_regsets[] = { · line 1413
Quoted 9 lines · Mark Brown
arch/arm64/kernel/ptrace.c +1
-1@@ -1392,7 +1407,7 @@ static const struct user_regset aarch64_regsets[] = { }, [REGSET_TLS] = { .core_note_type = NT_ARM_TLS, - .n = 1, + .n = 2, .size = sizeof(void *), .align = sizeof(void *), .regset_get = tls_get,Does this change confuse user-space? I presume an updated gdb would check the iov.len to figure out whether a new register is available but would existing debuggers complain of the new size of this regset?
gdb seems happy as far as I can see, it is possible something would be reusing the
read_iovfor repeated TLS read calls in a context where it was only pointing at a single u64 but I'm not sure how realistic that is given the idiom. I did do a search on sources.debian.net and didn't turn up anything that'd have problems.
Quoted 8 lines
Quoted 5 lines
If using this as an extensiblility mechanism is a concern we need to bear that in mind elsewhere, and for this it's either a case of providing another single register regset or trying to do a generic sysreg read/get (though that'd be another regset that's not idiomatic for the regset API).
Older GDB's assume a single register for
NT_ARM_TLS, so they will always fetch TPIDR. Newer GDB's will check the size and act accordingly.
That's fine. Looking at the ptrace_regset() implementation in Linux, it updates the user iov_len to what was actually copied. In this case it would be 1 (the minimum of the user iov_len and the regset n * size). So from the old gdb ABI perspective, it shouldn't notice a difference. An old gdb setting the TLS reg will also leave tpidr2_el0 unchanged.
An updated gdb running on an older kernel should be aware that even if it asked for two u64, it may only get one back and the user iov_len updated accordingly.
Signature · 1 line
Catalin
PATCH v4 4/4 kselftest/arm64: Add coverage of TPIDR2_EL0 ptrace interface
In reply to Mark BrownExtend the ptrace test support for NT_ARM_TLS to cover TPIDR2_EL0 - on systems that support SME the NT_ARM_TLS regset can be up to 2 elements long with the second element containing TPIDR2_EL0. On systems supporting SME we verify that this value can be read and written while on systems that do not support SME we verify correct truncation of reads and writes.
--- tools/testing/selftests/arm64/abi/ptrace.c | 82 +++++++++++++++++++++- 1 file changed, 79 insertions(+), 3 deletions(-)
tools/testing/selftests/arm64/abi/ptrace.c +79 -3
@@ -20,9 +20,9 @@#include "../../kselftest.h" -#define EXPECTED_TESTS 3 +#define EXPECTED_TESTS 7 -#define MAX_TPIDRS 1 +#define MAX_TPIDRS 2static bool have_sme (void ) {
@@ -34,7 +34,8 @@ static void test_tpidr(pid_t child)uint64_t read_val[MAX_TPIDRS ];uint64_t write_val[MAX_TPIDRS ];struct iovec read_iov, write_iov; -int ret; +bool test_tpidr2 = false; +int ret, i; read_iov.iov_base = read_val; write_iov.iov_base = write_val;
@@ -54,6 +55,81 @@ static void test_tpidr(pid_t child) ret =ptrace (PTRACE_GETREGSET , child,NT_ARM_TLS , &read_iov);ksft_test_result (ret ==0 && write_val[0 ] == read_val[0 ],"verify_tpidr_one\n" ); + +/* If we have TPIDR2 we should be able to read it */ + read_iov.iov_len =sizeof (read_val); + ret =ptrace (PTRACE_GETREGSET , child,NT_ARM_TLS , &read_iov); +if (ret ==0 ) { +/* If we have SME there should be two TPIDRs */ +if (read_iov.iov_len >=sizeof (read_val)) + test_tpidr2 = true; + +if (have_sme () && test_tpidr2) { +ksft_test_result (test_tpidr2,"count_tpidrs\n" ); + }else { +ksft_test_result (read_iov.iov_len %sizeof (uint64_t ) ==0 , +"count_tpidrs\n" ); + } + }else { +ksft_test_result_fail ("count_tpidrs\n" ); + } + +if (test_tpidr2) { +/* Try to write new values to all known TPIDRs... */ + write_iov.iov_len =sizeof (write_val); +for (i =0 ; i <MAX_TPIDRS ; i++) + write_val[i] = read_val[i] +1 ; + ret =ptrace (PTRACE_SETREGSET , child,NT_ARM_TLS , &write_iov); + +ksft_test_result (ret ==0 && + write_iov.iov_len ==sizeof (write_val), +"tpidr2_write\n" ); + +/* ...then read them back */ + read_iov.iov_len =sizeof (read_val); + ret =ptrace (PTRACE_GETREGSET , child,NT_ARM_TLS , &read_iov); + +if (have_sme ()) { +/* Should read back the written value */ +ksft_test_result (ret ==0 && + read_iov.iov_len >=sizeof (read_val) && +memcmp (read_val, write_val, +sizeof (read_val)) ==0 , +"tpidr2_read\n" ); + }else { +/* TPIDR2 should read as zero */ +ksft_test_result (ret ==0 && + read_iov.iov_len >=sizeof (read_val) && + read_val[0 ] == write_val[0 ] && + read_val[1 ] ==0 , +"tpidr2_read\n" ); + } + +/* Writing only TPIDR... */ + write_iov.iov_len =sizeof (uint64_t ); +memcpy (write_val, read_val,sizeof (read_val)); + write_val[0 ] +=1 ; + ret =ptrace (PTRACE_SETREGSET , child,NT_ARM_TLS , &write_iov); + +if (ret ==0 ) { +/* ...should leave TPIDR2 untouched */ + read_iov.iov_len =sizeof (read_val); + ret =ptrace (PTRACE_GETREGSET , child,NT_ARM_TLS , + &read_iov); + +ksft_test_result (ret ==0 && + read_iov.iov_len >=sizeof (read_val) && +memcmp (read_val, write_val, +sizeof (read_val)) ==0 , +"write_tpidr_only\n" ); + }else { +ksft_test_result_fail ("write_tpidr_only\n" ); + } + }else { +ksft_test_result_skip ("tpidr2_write\n" ); +ksft_test_result_skip ("tpidr2_read\n" ); +ksft_test_result_skip ("write_tpidr_only\n" ); + } }static int do_child (void )
Signature · 1 line
2.30.2
Quoted 8 lines · Mark Brown
When SME was initially merged we did not add support for
TPIDR2_EL0to the ptrace interface, creating difficulties for debuggers in accessing lazy save state for ZA. This series implements that support, extending the existingNT_ARM_TLSregset to support the register when available, and adds kselftest coverage for the existing and newNT_ARM_TLSfunctionality.[...]
Applied to arm64 (for-next/tpidr2_el0), thanks!
[1/4] kselftest/arm64: Add test coverage for NT_ARM_TLS
https://git.kernel.org/arm64/c/ecaf4d3f734f [2/4] arm64/ptrace: Document extension of NT_ARM_TLS to cover TPIDR2_EL0 https://git.kernel.org/arm64/c/f285da05c62a [3/4] arm64/ptrace: Support access to TPIDR2_EL0 https://git.kernel.org/arm64/c/0027d9c6c7b5 [4/4] kselftest/arm64: Add coverage of TPIDR2_EL0 ptrace interface https://git.kernel.org/arm64/c/2ddadec2206c
Signature · 1 line
Catalin