From: Luis Machado <luis.machado@arm.com>
To: Thiago Jung Bauermann <thiago.bauermann@linaro.org>,
gdb-patches@sourceware.org
Subject: Re: [PATCH 2/8] gdbserver: Add thread parameter to linux_get_auxv and linux_get_hwcap
Date: Tue, 20 Sep 2022 08:43:46 +0100 [thread overview]
Message-ID: <07552f1c-85bf-0a09-0656-15790248c036@arm.com> (raw)
In-Reply-To: <20220908064151.3959930-3-thiago.bauermann@linaro.org>
On 9/8/22 07:41, Thiago Jung Bauermann via Gdb-patches wrote:
> This patch doesn't change gdbserver behaviour, but after later changes are
> made it avoids a null pointer dereference when HWCAP needs to be obtained
> for a specific process while current_thread is nullptr.
>
> Fixing linux_read_auxv, linux_get_hwcap and linux_get_hwcap2 to take a
> thread_info parameter seems more correct than setting current_thread in one
> particular code path.
>
> Changes are propagated to allow passing the thread parameter through the
> call chain.
I suppose the effects will only be known when later patches are applied. But the auxv data is per-process, right?
It sounds a bit strange to pass a thread for the hwcap/auxv lookup. Is there some other reason to do that?
> ---
> gdbserver/linux-aarch64-low.cc | 6 +++---
> gdbserver/linux-arm-low.cc | 2 +-
> gdbserver/linux-low.cc | 20 +++++++++++---------
> gdbserver/linux-low.h | 10 +++++-----
> gdbserver/linux-ppc-low.cc | 6 +++---
> gdbserver/linux-s390-low.cc | 2 +-
> gdbserver/netbsd-low.cc | 4 ++--
> gdbserver/netbsd-low.h | 4 ++--
> gdbserver/server.cc | 2 +-
> gdbserver/target.cc | 4 ++--
> gdbserver/target.h | 4 ++--
> 11 files changed, 33 insertions(+), 31 deletions(-)
>
> diff --git a/gdbserver/linux-aarch64-low.cc b/gdbserver/linux-aarch64-low.cc
> index db5086962612..5d4d667dfa42 100644
> --- a/gdbserver/linux-aarch64-low.cc
> +++ b/gdbserver/linux-aarch64-low.cc
> @@ -826,9 +826,9 @@ aarch64_target::low_arch_setup ()
>
> features.vq = aarch64_sve_get_vq (tid);
> /* A-profile PAC is 64-bit only. */
> - features.pauth = linux_get_hwcap (8) & AARCH64_HWCAP_PACA;
> + features.pauth = linux_get_hwcap (current_thread, 8) & AARCH64_HWCAP_PACA;
> /* A-profile MTE is 64-bit only. */
> - features.mte = linux_get_hwcap2 (8) & HWCAP2_MTE;
> + features.mte = linux_get_hwcap2 (current_thread, 8) & HWCAP2_MTE;
> features.tls = true;
>
> current_process ()->tdesc = aarch64_linux_read_description (features);
> @@ -3299,7 +3299,7 @@ aarch64_target::supports_memory_tagging ()
> #endif
> }
>
> - return (linux_get_hwcap2 (8) & HWCAP2_MTE) != 0;
> + return (linux_get_hwcap2 (current_thread, 8) & HWCAP2_MTE) != 0;
> }
>
> bool
> diff --git a/gdbserver/linux-arm-low.cc b/gdbserver/linux-arm-low.cc
> index ab6209a3abc8..d40ce412fe1c 100644
> --- a/gdbserver/linux-arm-low.cc
> +++ b/gdbserver/linux-arm-low.cc
> @@ -958,7 +958,7 @@ get_next_pcs_syscall_next_pc (struct arm_get_next_pcs *self)
> static const struct target_desc *
> arm_read_description (void)
> {
> - unsigned long arm_hwcap = linux_get_hwcap (4);
> + unsigned long arm_hwcap = linux_get_hwcap (current_thread, 4);
>
> if (arm_hwcap & HWCAP_IWMMXT)
> return arm_linux_read_description (ARM_FP_TYPE_IWMMXT);
> diff --git a/gdbserver/linux-low.cc b/gdbserver/linux-low.cc
> index 2f71360d3bde..9034a1d8fabe 100644
> --- a/gdbserver/linux-low.cc
> +++ b/gdbserver/linux-low.cc
> @@ -5486,12 +5486,12 @@ linux_process_target::supports_read_auxv ()
> to debugger memory starting at MYADDR. */
>
> int
> -linux_process_target::read_auxv (CORE_ADDR offset, unsigned char *myaddr,
> - unsigned int len)
> +linux_process_target::read_auxv (const thread_info *thread, CORE_ADDR offset,
> + unsigned char *myaddr, unsigned int len)
> {
> char filename[PATH_MAX];
> int fd, n;
> - int pid = lwpid_of (current_thread);
> + int pid = lwpid_of (thread);
>
> xsnprintf (filename, sizeof filename, "/proc/%d/auxv", pid);
>
> @@ -6912,14 +6912,16 @@ linux_get_pc_64bit (struct regcache *regcache)
> /* See linux-low.h. */
>
> int
> -linux_get_auxv (int wordsize, CORE_ADDR match, CORE_ADDR *valp)
> +linux_get_auxv (const thread_info *thread, int wordsize, CORE_ADDR match,
> + CORE_ADDR *valp)
> {
> gdb_byte *data = (gdb_byte *) alloca (2 * wordsize);
> int offset = 0;
>
> gdb_assert (wordsize == 4 || wordsize == 8);
>
> - while (the_target->read_auxv (offset, data, 2 * wordsize) == 2 * wordsize)
> + while (the_target->read_auxv (thread, offset, data, 2 * wordsize)
> + == 2 * wordsize)
> {
> if (wordsize == 4)
> {
> @@ -6949,20 +6951,20 @@ linux_get_auxv (int wordsize, CORE_ADDR match, CORE_ADDR *valp)
> /* See linux-low.h. */
>
> CORE_ADDR
> -linux_get_hwcap (int wordsize)
> +linux_get_hwcap (const thread_info *thread, int wordsize)
> {
> CORE_ADDR hwcap = 0;
> - linux_get_auxv (wordsize, AT_HWCAP, &hwcap);
> + linux_get_auxv (thread, wordsize, AT_HWCAP, &hwcap);
> return hwcap;
> }
>
> /* See linux-low.h. */
>
> CORE_ADDR
> -linux_get_hwcap2 (int wordsize)
> +linux_get_hwcap2 (const thread_info *thread, int wordsize)
> {
> CORE_ADDR hwcap2 = 0;
> - linux_get_auxv (wordsize, AT_HWCAP2, &hwcap2);
> + linux_get_auxv (thread, wordsize, AT_HWCAP2, &hwcap2);
> return hwcap2;
> }
>
> diff --git a/gdbserver/linux-low.h b/gdbserver/linux-low.h
> index 79be31b8f729..bd1ec31dc5a4 100644
> --- a/gdbserver/linux-low.h
> +++ b/gdbserver/linux-low.h
> @@ -178,8 +178,8 @@ class linux_process_target : public process_stratum_target
>
> bool supports_read_auxv () override;
>
> - int read_auxv (CORE_ADDR offset, unsigned char *myaddr,
> - unsigned int len) override;
> + int read_auxv (const thread_info *thread, CORE_ADDR offset,
> + unsigned char *myaddr, unsigned int len) override;
>
> int insert_point (enum raw_bkpt_type type, CORE_ADDR addr,
> int size, raw_breakpoint *bp) override;
> @@ -944,17 +944,17 @@ extern int have_ptrace_getregset;
> *VALP and return 1. If not found or if there is an error, return
> 0. */
>
> -int linux_get_auxv (int wordsize, CORE_ADDR match,
> +int linux_get_auxv (const thread_info *thread, int wordsize, CORE_ADDR match,
> CORE_ADDR *valp);
>
> /* Fetch the AT_HWCAP entry from the auxv vector, where entries are length
> WORDSIZE. If no entry was found, return zero. */
>
> -CORE_ADDR linux_get_hwcap (int wordsize);
> +CORE_ADDR linux_get_hwcap (const thread_info *thread, int wordsize);
>
> /* Fetch the AT_HWCAP2 entry from the auxv vector, where entries are length
> WORDSIZE. If no entry was found, return zero. */
>
> -CORE_ADDR linux_get_hwcap2 (int wordsize);
> +CORE_ADDR linux_get_hwcap2 (const thread_info *thread, int wordsize);
>
> #endif /* GDBSERVER_LINUX_LOW_H */
> diff --git a/gdbserver/linux-ppc-low.cc b/gdbserver/linux-ppc-low.cc
> index 08824887003b..08608b2dcc38 100644
> --- a/gdbserver/linux-ppc-low.cc
> +++ b/gdbserver/linux-ppc-low.cc
> @@ -894,8 +894,8 @@ ppc_target::low_arch_setup ()
>
> /* The value of current_process ()->tdesc needs to be set for this
> call. */
> - ppc_hwcap = linux_get_hwcap (features.wordsize);
> - ppc_hwcap2 = linux_get_hwcap2 (features.wordsize);
> + ppc_hwcap = linux_get_hwcap (current_thread, features.wordsize);
> + ppc_hwcap2 = linux_get_hwcap2 (current_thread, features.wordsize);
>
> features.isa205 = ppc_linux_has_isa205 (ppc_hwcap);
>
> @@ -1097,7 +1097,7 @@ is_elfv2_inferior (void)
> const struct target_desc *tdesc = current_process ()->tdesc;
> int wordsize = register_size (tdesc, 0);
>
> - if (!linux_get_auxv (wordsize, AT_PHDR, &phdr))
> + if (!linux_get_auxv (current_thread, wordsize, AT_PHDR, &phdr))
> return def_res;
>
> /* Assume ELF header is at the beginning of the page where program headers
> diff --git a/gdbserver/linux-s390-low.cc b/gdbserver/linux-s390-low.cc
> index 5adc28070574..0d01dffa64a4 100644
> --- a/gdbserver/linux-s390-low.cc
> +++ b/gdbserver/linux-s390-low.cc
> @@ -592,7 +592,7 @@ s390_target::low_arch_setup ()
> /* Determine word size and HWCAP. */
> int pid = pid_of (current_thread);
> int wordsize = s390_get_wordsize (pid);
> - unsigned long hwcap = linux_get_hwcap (wordsize);
> + unsigned long hwcap = linux_get_hwcap (current_thread, wordsize);
>
> /* Check whether the kernel supports extra register sets. */
> int have_regset_last_break
> diff --git a/gdbserver/netbsd-low.cc b/gdbserver/netbsd-low.cc
> index f05bcd4e173f..71ac61e9dfbd 100644
> --- a/gdbserver/netbsd-low.cc
> +++ b/gdbserver/netbsd-low.cc
> @@ -581,10 +581,10 @@ netbsd_read_auxv(pid_t pid, void *offs, void *addr, size_t len)
> to debugger memory starting at MYADDR. */
>
> int
> -netbsd_process_target::read_auxv (CORE_ADDR offset,
> +netbsd_process_target::read_auxv (const thread_info *thread, CORE_ADDR offset,
> unsigned char *myaddr, unsigned int len)
> {
> - pid_t pid = pid_of (current_thread);
> + pid_t pid = pid_of (thread);
>
> return netbsd_read_auxv (pid, (void *) (intptr_t) offset, myaddr, len);
> }
> diff --git a/gdbserver/netbsd-low.h b/gdbserver/netbsd-low.h
> index 06b4fd613571..506329251eaf 100644
> --- a/gdbserver/netbsd-low.h
> +++ b/gdbserver/netbsd-low.h
> @@ -77,8 +77,8 @@ class netbsd_process_target : public process_stratum_target
>
> bool supports_read_auxv () override;
>
> - int read_auxv (CORE_ADDR offset, unsigned char *myaddr,
> - unsigned int len) override;
> + int read_auxv (const thread_info *thread, CORE_ADDR offset,
> + unsigned char *myaddr, unsigned int len) override;
>
> bool supports_hardware_single_step () override;
>
> diff --git a/gdbserver/server.cc b/gdbserver/server.cc
> index c619206d5d2d..2d5cf6e8d8de 100644
> --- a/gdbserver/server.cc
> +++ b/gdbserver/server.cc
> @@ -1443,7 +1443,7 @@ handle_qxfer_auxv (const char *annex,
> if (annex[0] != '\0' || current_thread == NULL)
> return -1;
>
> - return the_target->read_auxv (offset, readbuf, len);
> + return the_target->read_auxv (current_thread, offset, readbuf, len);
> }
>
> /* Handle qXfer:exec-file:read. */
> diff --git a/gdbserver/target.cc b/gdbserver/target.cc
> index adcfe6e7bcc7..7ee18bde8635 100644
> --- a/gdbserver/target.cc
> +++ b/gdbserver/target.cc
> @@ -344,8 +344,8 @@ process_stratum_target::supports_read_auxv ()
> }
>
> int
> -process_stratum_target::read_auxv (CORE_ADDR offset, unsigned char *myaddr,
> - unsigned int len)
> +process_stratum_target::read_auxv (const thread_info *thread, CORE_ADDR offset,
> + unsigned char *myaddr, unsigned int len)
> {
> gdb_assert_not_reached ("target op read_auxv not supported");
> }
> diff --git a/gdbserver/target.h b/gdbserver/target.h
> index 6c536a307786..cd6b2193a228 100644
> --- a/gdbserver/target.h
> +++ b/gdbserver/target.h
> @@ -175,8 +175,8 @@ class process_stratum_target
> /* Read auxiliary vector data from the inferior process.
>
> Read LEN bytes at OFFSET into a buffer at MYADDR. */
> - virtual int read_auxv (CORE_ADDR offset, unsigned char *myaddr,
> - unsigned int len);
> + virtual int read_auxv (const thread_info *thread, CORE_ADDR offset,
> + unsigned char *myaddr, unsigned int len);
>
> /* Returns true if GDB Z breakpoint type TYPE is supported, false
> otherwise. The type is coded as follows:
next prev parent reply other threads:[~2022-09-20 7:44 UTC|newest]
Thread overview: 25+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-09-08 6:41 [PATCH 0/8] gdbserver improvements for AArch64 SVE support Thiago Jung Bauermann
2022-09-08 6:41 ` [PATCH 1/8] gdbserver: Add asserts in register_size and register_data functions Thiago Jung Bauermann
2022-09-20 7:36 ` Luis Machado
2022-11-26 1:36 ` Thiago Jung Bauermann
2022-09-08 6:41 ` [PATCH 2/8] gdbserver: Add thread parameter to linux_get_auxv and linux_get_hwcap Thiago Jung Bauermann
2022-09-20 7:43 ` Luis Machado [this message]
2022-11-26 1:37 ` Thiago Jung Bauermann
2022-09-08 6:41 ` [PATCH 3/8] gdb, gdbserver/aarch64-linux: Allow aarch64_sve_get_vq to return error Thiago Jung Bauermann
2022-09-20 8:07 ` Luis Machado
2022-11-26 1:42 ` Thiago Jung Bauermann
2022-09-08 6:41 ` [PATCH 4/8] gdbserver/linux-aarch64: Factor out function to get aarch64_features Thiago Jung Bauermann
2022-09-20 8:08 ` Luis Machado
2022-11-26 1:44 ` Thiago Jung Bauermann
2022-09-08 6:41 ` [PATCH 5/8] gdbserver: Move target description from being per-process to being per-thread Thiago Jung Bauermann
2022-09-20 11:21 ` Luis Machado
2022-11-26 1:47 ` Thiago Jung Bauermann
2022-09-08 6:41 ` [PATCH 6/8] gdbserver/linux-aarch64: When inferior stops, update its target description Thiago Jung Bauermann
2022-09-20 8:48 ` Luis Machado
2022-11-26 1:49 ` Thiago Jung Bauermann
2022-09-08 6:41 ` [PATCH 7/8] gdb/aarch64: Factor out most of the thread_architecture method Thiago Jung Bauermann
2022-09-20 8:52 ` Luis Machado
2022-11-26 1:49 ` Thiago Jung Bauermann
2022-09-08 6:41 ` [PATCH 8/8] gdb/aarch64: When remote debugging detect changes in the target description Thiago Jung Bauermann
2022-09-20 8:59 ` Luis Machado
2022-11-26 1:50 ` Thiago Jung Bauermann
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=07552f1c-85bf-0a09-0656-15790248c036@arm.com \
--to=luis.machado@arm.com \
--cc=gdb-patches@sourceware.org \
--cc=thiago.bauermann@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 read-only IMAP folder(s) and NNTP newsgroup(s).