Re: [PATCH v2 19/28] arm64/sve: ptrace and ELF coredump support

Dave Martin <[email protected]>
Newsgroups gmane.linux.kernel.cross-arch,gmane.linux.ports.arm.kernel,gmane.comp.lib.glibc.alpha,gmane.comp.gdb.devel,gmane.comp.emulators.kvm.arm.devel
Message-ID <[email protected]>
On Wed, Sep 06, 2017 at 04:21:50PM +0000, Okamoto, Takayuki wrote:
> Hi Dave,
> 
> I am an engineer of the postK computer from Fujitsu.
> 
> When I tried to read "max_vl" by ptrace with this patch on our local SVE 
> simulator, it was read as zero.
> I think the cause of this incident is that "max_vl" is set as "header->vl" 
> only on warning case in sve_init_header_from_task().
> "max_vl" should be set up also on normal case, like the following patch.
> 
> 
> --- a/arch/arm64/kernel/ptrace.c
> +++ b/arch/arm64/kernel/ptrace.c
> @@ -755,6 +755,8 @@ static void sve_init_header_from_task(struct user_sve_header *header,
> 
>         if (WARN_ON(!sve_vl_valid(sve_max_vl)))
>                 header->max_vl = header->vl;
> +       else
> +               header->max_vl = sve_max_vl;
> 
>         header->size = SVE_PT_SIZE(vq, header->flags);
>         header->max_size = SVE_PT_SIZE(sve_vq_from_vl(header->max_vl),

Hi, thanks for reporting this.

It looks like a refactoring mistake I made while removing BUG_ON()s,
which I missed in my testing.

Your fix looks correct and seems to work.  For stylistic reasons, I may
write it like this instead, but the effect should be the same:

	header->max_vl = sve_max_vl;
	if (WARN_ON(!sve_vl_valid(sve_max_vl))
		header->max_vl = header->vl;

Cheers
---Dave

> 
> 
> Best regards,
> Takayuki Okamoto
> 
> -----Original Message-----
> From: [email protected] [mailto:[email protected]] On Behalf Of Dave Martin
> Sent: Friday, September 1, 2017 2:01 AM
> To: [email protected]
> Cc: Catalin Marinas <[email protected]>; Will Deacon <[email protected]>; Ard Biesheuvel <[email protected]>; Alex Bennée <[email protected]>; Szabolcs Nagy <[email protected]>; Richard Sandiford <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]; Alan Hayward <[email protected]>; Yao Qi <[email protected]>; Oleg Nesterov <[email protected]>; Alexander Viro <[email protected]>
> Subject: [PATCH v2 19/28] arm64/sve: ptrace and ELF coredump support
> 

[...]

> @@ -702,6 +737,210 @@ static int system_call_set(struct task_struct *target,
>  	return ret;
>  }
>  
> +#ifdef CONFIG_ARM64_SVE
> +
> +static void sve_init_header_from_task(struct user_sve_header *header,
> +				      struct task_struct *target)
> +{
> +	unsigned int vq;
> +
> +	memset(header, 0, sizeof(*header));
> +
> +	header->flags = test_tsk_thread_flag(target, TIF_SVE) ?
> +		SVE_PT_REGS_SVE : SVE_PT_REGS_FPSIMD;
> +	if (test_tsk_thread_flag(target, TIF_SVE_VL_INHERIT))
> +		header->flags |= SVE_PT_VL_INHERIT;
> +
> +	header->vl = target->thread.sve_vl;
> +	vq = sve_vq_from_vl(header->vl);
> +
> +	if (WARN_ON(!sve_vl_valid(sve_max_vl)))
> +		header->max_vl = header->vl;
> +
> +	header->size = SVE_PT_SIZE(vq, header->flags);
> +	header->max_size = SVE_PT_SIZE(sve_vq_from_vl(header->max_vl),
> +				      SVE_PT_REGS_SVE);
> +}

[...]
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.