linux-kernel - Re: [PATCH V2 3/7] iommu/vt-d: Support Enhanced Command Interface

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]

Date:   Thu, 19 Jan 2023 16:55:45 +0800
From:   Baolu Lu <baolu.lu@...ux.intel.com>
To:     kan.liang@...ux.intel.com, joro@...tes.org, will@...nel.org,
        dwmw2@...radead.org, robin.murphy@....com, robert.moore@...el.com,
        rafael.j.wysocki@...el.com, lenb@...nel.org, iommu@...ts.linux.dev,
        linux-kernel@...r.kernel.org
Cc:     baolu.lu@...ux.intel.com
Subject: Re: [PATCH V2 3/7] iommu/vt-d: Support Enhanced Command Interface

On 2023/1/19 4:50, kan.liang@...ux.intel.com wrote:
> +#ifdef CONFIG_INTEL_IOMMU
> +#define ecmd_get_status_code(res)	((res & 0xff) >> 1)
> +
> +/*
> + * Function to submit a command to the enhanced command interface. The
> + * valid enhanced command descriptions are defined in Table 47 of the
> + * VT-d spec. The VT-d hardware implementation may support some but not
> + * all commands, which can be determined by checking the Enhanced
> + * Command Capability Register.
> + *
> + * Return values:
> + *  - 0: Command successful without any error;
> + *  - Negative: software error value;
> + *  - Nonzero positive: failure status code defined in Table 48.
> + */
> +int ecmd_submit_sync(struct intel_iommu *iommu, u8 ecmd, u64 oa, u64 ob)
> +{
> +	unsigned long flags;
> +	u64 res;
> +	int ret;
> +
> +	if (!cap_ecmds(iommu->cap))
> +		return -ENODEV;
> +
> +	raw_spin_lock_irqsave(&iommu->register_lock, flags);
> +
> +	res = dmar_readq(iommu->reg + DMAR_ECRSP_REG);
> +	if (res & DMA_ECMD_ECRSP_IP) {
> +		ret = -EBUSY;
> +		goto err;
> +	}
> +
> +	/*
> +	 * Unconditionally write the operand B, because
> +	 * - There is no side effect if an ecmd doesn't require an
> +	 *   operand B, but we set the register to some value.
> +	 * - It's not invoked in any critical path. The extra MMIO
> +	 *   write doesn't bring any performance concerns.
> +	 */
> +	dmar_writeq(iommu->reg + DMAR_ECEO_REG, ob);
> +	dmar_writeq(iommu->reg + DMAR_ECMD_REG, ecmd | (oa << DMA_ECMD_OA_SHIFT));
> +
> +	IOMMU_WAIT_OP(iommu, DMAR_ECRSP_REG, dmar_readq,
> +		      !(res & DMA_ECMD_ECRSP_IP), res);
> +
> +	if (res & DMA_ECMD_ECRSP_IP) {
> +		ret = -ETIMEDOUT;
> +		goto err;
> +	}
> +
> +	ret = ecmd_get_status_code(res);
> +err:
> +	raw_spin_unlock_irqrestore(&iommu->register_lock, flags);
> +
> +	return ret;
> +}
> +#endif /* CONFIG_INTEL_IOMMU */

Can we remove the "#ifdef CONFIG_INTEL_IOMMU"? Or if this is currently
only intel-iommu specific, how about moving it to drivers/iommu/intel
/iommu.c?

--
Best regards,
baolu