linux-kernel - Re: [PATCH 2/3] habanalabs: trace MMU map/unmap page

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: <20220811095046.0f303eac@gandalf.local.home>
Date:   Thu, 11 Aug 2022 09:50:46 -0400
From:   Steven Rostedt <rostedt@...dmis.org>
To:     Oded Gabbay <ogabbay@...nel.org>
Cc:     linux-kernel@...r.kernel.org, gregkh@...uxfoundation.org,
        Ohad Sharabi <osharabi@...ana.ai>
Subject: Re: [PATCH 2/3] habanalabs: trace MMU map/unmap page

On Thu, 11 Aug 2022 14:39:52 +0300
Oded Gabbay <ogabbay@...nel.org> wrote:

> @@ -259,7 +261,12 @@ int hl_mmu_unmap_page(struct hl_ctx *ctx, u64 virt_addr, u32 page_size, bool flu
>  	if (flush_pte)
>  		mmu_funcs->flush(ctx);
>  
> -	return rc;
> +	if (rc)
> +		return rc;

If you want to avoid the condition if tracing is disabled (for a
micro-optimization) you can instead do:

	if (trace_habanalabs_mmu_unmap_enabled() && !rc)
		trace_habanalabs_mmu_unmap(hdev->dev, virt_addr, 0, page_size, flush_pte);

	return rc;

This way the above trace_habanalabs_mmu_unmap_enabled() is a static_branch,
and will become a nop falling into the "return rc", and when the tracepoint
is enabled, it will jump to a compare of "!rc".

-- Steve


> +
> +	trace_habanalabs_mmu_unmap(hdev->dev, virt_addr, 0, page_size, flush_pte);
> +
> +	return 0;
>  }
>