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  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]
Date:   Tue, 20 Jul 2021 15:54:37 +0200
From:   Christoph Hellwig <hch@....de>
To:     Tianyu Lan <ltykernel@...il.com>
Cc:     kys@...rosoft.com, haiyangz@...rosoft.com, sthemmin@...rosoft.com,
        wei.liu@...nel.org, decui@...rosoft.com, tglx@...utronix.de,
        mingo@...hat.com, bp@...en8.de, x86@...nel.org, hpa@...or.com,
        dave.hansen@...ux.intel.com, luto@...nel.org, peterz@...radead.org,
        konrad.wilk@...cle.com, boris.ostrovsky@...cle.com,
        jgross@...e.com, sstabellini@...nel.org, joro@...tes.org,
        will@...nel.org, davem@...emloft.net, kuba@...nel.org,
        jejb@...ux.ibm.com, martin.petersen@...cle.com, arnd@...db.de,
        hch@....de, m.szyprowski@...sung.com, robin.murphy@....com,
        kirill.shutemov@...ux.intel.com, akpm@...ux-foundation.org,
        rppt@...nel.org, Tianyu.Lan@...rosoft.com, thomas.lendacky@....com,
        ardb@...nel.org, robh@...nel.org, nramas@...ux.microsoft.com,
        pgonda@...gle.com, martin.b.radev@...il.com, david@...hat.com,
        krish.sadhukhan@...cle.com, saravanand@...com,
        xen-devel@...ts.xenproject.org, keescook@...omium.org,
        rientjes@...gle.com, hannes@...xchg.org,
        michael.h.kelley@...rosoft.com, iommu@...ts.linux-foundation.org,
        linux-arch@...r.kernel.org, linux-hyperv@...r.kernel.org,
        linux-kernel@...r.kernel.org, linux-scsi@...r.kernel.org,
        netdev@...r.kernel.org, vkuznets@...hat.com, brijesh.singh@....com,
        anparri@...rosoft.com
Subject: Re: [Resend RFC PATCH V4 09/13] x86/Swiotlb/HV: Add Swiotlb bounce
 buffer remap function for HV IVM


Please split the swiotlb changes into a separate patch from the
consumer.

>  }
> +
> +/*
> + * hv_map_memory - map memory to extra space in the AMD SEV-SNP Isolation VM.
> + */
> +unsigned long hv_map_memory(unsigned long addr, unsigned long size)
> +{
> +	unsigned long *pfns = kcalloc(size / HV_HYP_PAGE_SIZE,
> +				      sizeof(unsigned long),
> +		       GFP_KERNEL);
> +	unsigned long vaddr;
> +	int i;
> +
> +	if (!pfns)
> +		return (unsigned long)NULL;
> +
> +	for (i = 0; i < size / HV_HYP_PAGE_SIZE; i++)
> +		pfns[i] = virt_to_hvpfn((void *)addr + i * HV_HYP_PAGE_SIZE) +
> +			(ms_hyperv.shared_gpa_boundary >> HV_HYP_PAGE_SHIFT);
> +
> +	vaddr = (unsigned long)vmap_pfn(pfns, size / HV_HYP_PAGE_SIZE,
> +					PAGE_KERNEL_IO);
> +	kfree(pfns);
> +
> +	return vaddr;

This seems to miss a 'select VMAP_PFN'.  But more importantly I don't
think this actually works.  Various DMA APIs do expect a struct page
backing, so how is this going to work with say dma_mmap_attrs or
dma_get_sgtable_attrs?

> +static unsigned long __map_memory(unsigned long addr, unsigned long size)
> +{
> +	if (hv_is_isolation_supported())
> +		return hv_map_memory(addr, size);
> +
> +	return addr;
> +}
> +
> +static void __unmap_memory(unsigned long addr)
> +{
> +	if (hv_is_isolation_supported())
> +		hv_unmap_memory(addr);
> +}
> +
> +unsigned long set_memory_decrypted_map(unsigned long addr, unsigned long size)
> +{
> +	if (__set_memory_enc_dec(addr, size / PAGE_SIZE, false))
> +		return (unsigned long)NULL;
> +
> +	return __map_memory(addr, size);
> +}
> +
> +int set_memory_encrypted_unmap(unsigned long addr, unsigned long size)
> +{
> +	__unmap_memory(addr);
> +	return __set_memory_enc_dec(addr, size / PAGE_SIZE, true);
> +}

Why this obsfucation into all kinds of strange helpers?  Also I think
we want an ops vectors (or alternative calls) instead of the random
if checks here.

> + * @vstart:	The virtual start address of the swiotlb memory pool. The swiotlb
> + *		memory pool may be remapped in the memory encrypted case and store

Normall we'd call this vaddr or cpu_addr.

> -	set_memory_decrypted((unsigned long)vaddr, bytes >> PAGE_SHIFT);
> -	memset(vaddr, 0, bytes);
> +	mem->vstart = (void *)set_memory_decrypted_map((unsigned long)vaddr, bytes);

Please always pass kernel virtual addresses as pointers.

And I think these APIs might need better names, e.g.

arch_dma_map_decrypted and arch_dma_unmap_decrypted.

Also these will need fallback versions for non-x86 architectures that
currently use memory encryption.

Powered by blists - more mailing lists