linux-kernel - Re: [PATCH 1/6] arm64/vdso: use the fault callback to map vvar pages

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20200618072705.GA334958@gmail.com>
Date:   Thu, 18 Jun 2020 00:27:05 -0700
From:   Andrei Vagin <avagin@...il.com>
To:     Mark Rutland <mark.rutland@....com>
Cc:     linux-arm-kernel@...ts.infradead.org,
        Catalin Marinas <catalin.marinas@....com>,
        Will Deacon <will@...nel.org>, linux-kernel@...r.kernel.org,
        Vincenzo Frascino <vincenzo.frascino@....com>,
        Thomas Gleixner <tglx@...utronix.de>,
        Dmitry Safonov <dima@...sta.com>
Subject: Re: [PATCH 1/6] arm64/vdso: use the fault callback to map vvar pages

On Tue, Jun 16, 2020 at 11:59:20AM +0100, Mark Rutland wrote:
> On Tue, Jun 16, 2020 at 12:55:40AM -0700, Andrei Vagin wrote:
> > This is required to support time namespaces where a time namespace data
> > page is different for each namespace.
> 
> Can you please give a bit more of an introduction to the changes here?
> As-is, this doesn't give reviewers the necessary context to understand
> the change, nor does it justify it.
> 
> Ideally, a commit message for this should look something like:

Mark, thank you for the review. I will write more details message and
fix this patch according with your other comments.

> 
> | Currently the vdso has no awareness of time namespaces, which may
> | apply distinct offsets to processes in different namespaces. To handle
> | this within the vdso, we'll need to expose a per-namespace data page.
> |
> | As a preparatory step, this patch separates the vdso data page from
> | the code pages, and has it faulted in via its own fault callback.
> | Subsquent patches will extend this to support distinct pages per time
> | namespace.
> 
> Otherwise, I have a few minor comments below.
> 
> > Reviewed-by: Vincenzo Frascino <vincenzo.frascino@....com>
> > Reviewed-by: Dmitry Safonov <dima@...sta.com>
> > Signed-off-by: Andrei Vagin <avagin@...il.com>
> > ---
> >  arch/arm64/kernel/vdso.c | 24 +++++++++++++++---------
> >  1 file changed, 15 insertions(+), 9 deletions(-)
> > 
> > diff --git a/arch/arm64/kernel/vdso.c b/arch/arm64/kernel/vdso.c
> > index 4e016574bd91..b0aec4e8c9b4 100644
> > --- a/arch/arm64/kernel/vdso.c
> > +++ b/arch/arm64/kernel/vdso.c
> > @@ -108,28 +108,32 @@ static int __vdso_init(enum vdso_abi abi)
> >  			PAGE_SHIFT;
> >  
> >  	/* Allocate the vDSO pagelist, plus a page for the data. */
> > -	vdso_pagelist = kcalloc(vdso_info[abi].vdso_pages + 1,
> > +	vdso_pagelist = kcalloc(vdso_info[abi].vdso_pages,
> >  				sizeof(struct page *),
> >  				GFP_KERNEL);
> 
> The comment above is now stale. Can you please update it, or (event
> better) delete it entirely?

I will delete it.

> 
> >  	if (vdso_pagelist == NULL)
> >  		return -ENOMEM;
> >  
> > -	/* Grab the vDSO data page. */
> > -	vdso_pagelist[0] = phys_to_page(__pa_symbol(vdso_data));
> > -
> > -
> >  	/* Grab the vDSO code pages. */
> >  	pfn = sym_to_pfn(vdso_info[abi].vdso_code_start);
> >  
> >  	for (i = 0; i < vdso_info[abi].vdso_pages; i++)
> > -		vdso_pagelist[i + 1] = pfn_to_page(pfn + i);
> > +		vdso_pagelist[i] = pfn_to_page(pfn + i);
> >  
> > -	vdso_info[abi].dm->pages = &vdso_pagelist[0];
> > -	vdso_info[abi].cm->pages = &vdso_pagelist[1];
> > +	vdso_info[abi].cm->pages = vdso_pagelist;
> >  
> >  	return 0;
> >  }
> >  
> > +static vm_fault_t vvar_fault(const struct vm_special_mapping *sm,
> > +			     struct vm_area_struct *vma, struct vm_fault *vmf)
> > +{
> > +	if (vmf->pgoff == 0)
> > +		return vmf_insert_pfn(vma, vmf->address,
> > +				sym_to_pfn(vdso_data));
> > +	return VM_FAULT_SIGBUS;
> > +}
> 
> This might look better as:

In the next patch, this function will handle more pages and this form
will look better.

> 
> |	if (vmf->pgoff != 0)
> |		return VM_FAULT_SIGBUS;
> |	
> |	return vmf_insert_pfn(vma, vmf->address, sym_to_pfn(vdso_data));
> 
> > +
> >  static int __setup_additional_pages(enum vdso_abi abi,
> >  				    struct mm_struct *mm,
> >  				    struct linux_binprm *bprm,
> > @@ -150,7 +154,7 @@ static int __setup_additional_pages(enum vdso_abi abi,
> >  	}
> >  
> >  	ret = _install_special_mapping(mm, vdso_base, PAGE_SIZE,
> > -				       VM_READ|VM_MAYREAD,
> > +				       VM_READ|VM_MAYREAD|VM_PFNMAP,
> 
> This change needs to be explained in the commit message. WHy is it
> necessary, and why only so for the data page?

I will update the commit message. VM_PFNMAP is required to handle faults
from a vma fault callback. In this case, it is vvar_fault.

> 
> Thanks,
> Mark.
> 
> >  				       vdso_info[abi].dm);
> >  	if (IS_ERR(ret))
> >  		goto up_fail;
> > @@ -209,6 +213,7 @@ static struct vm_special_mapping aarch32_vdso_maps[] = {
> >  #ifdef CONFIG_COMPAT_VDSO
> >  	[AA32_MAP_VVAR] = {
> >  		.name = "[vvar]",
> > +		.fault = vvar_fault,
> >  	},
> >  	[AA32_MAP_VDSO] = {
> >  		.name = "[vdso]",
> > @@ -376,6 +381,7 @@ enum aarch64_map {
> >  static struct vm_special_mapping aarch64_vdso_maps[] __ro_after_init = {
> >  	[AA64_MAP_VVAR] = {
> >  		.name	= "[vvar]",
> > +		.fault = vvar_fault,
> >  	},
> >  	[AA64_MAP_VDSO] = {
> >  		.name	= "[vdso]",
> > -- 
> > 2.24.1
> >