linux-kernel - RE: [PATCH 1/6] cxl/trace: Pass uuid explicitly to event traces

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: <657263c249c4c_45e0129471@dwillia2-xfh.jf.intel.com.notmuch>
Date:   Thu, 7 Dec 2023 16:30:58 -0800
From:   Dan Williams <dan.j.williams@...el.com>
To:     Ira Weiny <ira.weiny@...el.com>,
        Dan Williams <dan.j.williams@...el.com>,
        Jonathan Cameron <jonathan.cameron@...wei.com>,
        Smita Koralahalli <Smita.KoralahalliChannabasappa@....com>,
        Shiju Jose <shiju.jose@...wei.com>
CC:     Yazen Ghannam <yazen.ghannam@....com>,
        Davidlohr Bueso <dave@...olabs.net>,
        Dave Jiang <dave.jiang@...el.com>,
        Alison Schofield <alison.schofield@...el.com>,
        Vishal Verma <vishal.l.verma@...el.com>,
        "Ard Biesheuvel" <ardb@...nel.org>, <linux-efi@...r.kernel.org>,
        <linux-kernel@...r.kernel.org>, <linux-cxl@...r.kernel.org>,
        Ira Weiny <ira.weiny@...el.com>
Subject: RE: [PATCH 1/6] cxl/trace: Pass uuid explicitly to event traces

Ira Weiny wrote:
> CPER CXL events do not have a UUID associated with them.  It is
> desirable to share event structures between the CPER CXL event and the
> CXL event log events.

This parses strange to me, maybe it is trying to capture too many
changes in too few sentences? Something like this instead?

"CXL CPER events are identified by the CPER Section Type GUID. The GUID
correlates with the CXL UUID for the event record. It turns out that a
CXL CPER record is a strict subset of the CXL event record, only the
UUID header field is chopped.

In order to unify handling between native and CPER flavors of CXL
events, prepare the code for the UUID to be passed in rather than
inferred from the record itself.

Later patches update the passed in record to only refer to the common
data between the formats."

> 
> Pass the UUID explicitly to each trace event to be able to remove the
> UUID from the event structures.
> 
> Originally it was desirable to remove the UUID from the well known event
> because the UUID value was redundant.  However, the trace API was
> already in place.[1]
> 
> [1] https://lore.kernel.org/all/36f2d12934d64a278f2c0313cbd01abc@huawei.com/
> 
> Signed-off-by: Ira Weiny <ira.weiny@...el.com>
> ---
>  drivers/cxl/core/mbox.c  |  8 ++++----
>  drivers/cxl/core/trace.h | 32 ++++++++++++++++----------------
>  2 files changed, 20 insertions(+), 20 deletions(-)
> 
> diff --git a/drivers/cxl/core/mbox.c b/drivers/cxl/core/mbox.c
> index 36270dcfb42e..00f429c440df 100644
> --- a/drivers/cxl/core/mbox.c
> +++ b/drivers/cxl/core/mbox.c
> @@ -870,19 +870,19 @@ static void cxl_event_trace_record(const struct cxl_memdev *cxlmd,
>  		struct cxl_event_gen_media *rec =
>  				(struct cxl_event_gen_media *)record;
>  
> -		trace_cxl_general_media(cxlmd, type, rec);
> +		trace_cxl_general_media(cxlmd, type, id, rec);
>  	} else if (uuid_equal(id, &dram_event_uuid)) {
>  		struct cxl_event_dram *rec = (struct cxl_event_dram *)record;
>  
> -		trace_cxl_dram(cxlmd, type, rec);
> +		trace_cxl_dram(cxlmd, type, id, rec);
>  	} else if (uuid_equal(id, &mem_mod_event_uuid)) {
>  		struct cxl_event_mem_module *rec =
>  				(struct cxl_event_mem_module *)record;
>  
> -		trace_cxl_memory_module(cxlmd, type, rec);
> +		trace_cxl_memory_module(cxlmd, type, id, rec);
>  	} else {
>  		/* For unknown record types print just the header */
> -		trace_cxl_generic_event(cxlmd, type, record);
> +		trace_cxl_generic_event(cxlmd, type, id, record);
>  	}
>  }
>  
> diff --git a/drivers/cxl/core/trace.h b/drivers/cxl/core/trace.h
> index a0b5819bc70b..2aef185f4cd0 100644
> --- a/drivers/cxl/core/trace.h
> +++ b/drivers/cxl/core/trace.h
> @@ -189,7 +189,7 @@ TRACE_EVENT(cxl_overflow,
>  	__string(memdev, dev_name(&cxlmd->dev))			\
>  	__string(host, dev_name(cxlmd->dev.parent))		\
>  	__field(int, log)					\
> -	__field_struct(uuid_t, hdr_uuid)			\
> +	__field_struct(uuid_t, uuid)				\

Hmm, is it too late to make this rename? I.e. would a script that is
looking for "hdr_uuid" in one kernel be broken by the rename of the
field to "uuid" in the next?

>  	__field(u64, serial)					\
>  	__field(u32, hdr_flags)					\
>  	__field(u16, hdr_handle)				\
> @@ -198,12 +198,12 @@ TRACE_EVENT(cxl_overflow,
>  	__field(u8, hdr_length)					\
>  	__field(u8, hdr_maint_op_class)
>  
> -#define CXL_EVT_TP_fast_assign(cxlmd, l, hdr)					\
> +#define CXL_EVT_TP_fast_assign(cxlmd, l, uuid, hdr)				\
>  	__assign_str(memdev, dev_name(&(cxlmd)->dev));				\
>  	__assign_str(host, dev_name((cxlmd)->dev.parent));			\
>  	__entry->log = (l);							\
>  	__entry->serial = (cxlmd)->cxlds->serial;				\
> -	memcpy(&__entry->hdr_uuid, &(hdr).id, sizeof(uuid_t));			\
> +	memcpy(&__entry->uuid, (uuid), sizeof(uuid_t));				\

Per above this would be:

	memcpy(&__entry->hdr_uuid, (uuid), sizeof(uuid_t));				\

>  	__entry->hdr_length = (hdr).length;					\
>  	__entry->hdr_flags = get_unaligned_le24((hdr).flags);			\
>  	__entry->hdr_handle = le16_to_cpu((hdr).handle);			\
> @@ -217,7 +217,7 @@ TRACE_EVENT(cxl_overflow,
>  		"maint_op_class=%u : " fmt,					\
>  		__get_str(memdev), __get_str(host), __entry->serial,		\
>  		cxl_event_log_type_str(__entry->log),				\
> -		__entry->hdr_timestamp, &__entry->hdr_uuid, __entry->hdr_length,\
> +		__entry->hdr_timestamp, &__entry->uuid, __entry->hdr_length,	\

...and this change could be dropped.

Other than that, this looks good to me.