mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Xin Li <xin@zytor.com>
To: Nikolay Borisov <nik.borisov@suse.com>,
	linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org
Cc: tglx@linutronix.de, mingo@redhat.com, bp@alien8.de,
	dave.hansen@linux.intel.com, x86@kernel.org, hpa@zytor.com,
	will@kernel.org, peterz@infradead.org, akpm@linux-foundation.org,
	acme@kernel.org, namhyung@kernel.org, brgerst@gmail.com
Subject: Re: [PATCH v3 2/4] x86/cpufeatures: Generate a feature mask header based on build config
Date: Thu, 27 Jun 2024 09:45:34 -0700	[thread overview]
Message-ID: <e84aa8f0-3885-452c-9250-c5910d869263@zytor.com> (raw)
In-Reply-To: <5aeaaecf-0eb6-4934-b14e-2285eff6cc62@suse.com>

On 6/26/2024 4:36 AM, Nikolay Borisov wrote:
> 
> 
> On 22.06.24 г. 20:14 ч., Xin Li (Intel) wrote:
>> From: "H. Peter Anvin (Intel)" <hpa@zytor.com>
>>
>> Introduce an AWK script to auto-generate a header with required and
>> disabled feature masks based on <asm/cpufeatures.h> and current build
>> config. Thus for any CPU feature with a build config, e.g., X86_FRED,
>> simply add
>>
>> config X86_DISABLED_FEATURE_FRED
>>     def_bool y
>>     depends on !X86_FRED
>>
>> to arch/x86/Kconfig.cpufeatures, instead of adding a conditional CPU
>> feature disable flag, e.g., DISABLE_FRED.
>>
>> Lastly the generated required and disabled feature masks will be added
>> to their corresponding feature masks for this particular compile-time
>> configuration.
>>
>> [ Xin: build integration improvements ]
>>
>> Signed-off-by: H. Peter Anvin (Intel) <hpa@zytor.com>
>> Signed-off-by: Xin Li (Intel) <xin@zytor.com>
> 
> Overall LGTM, some minor points below.
> 
> 
> Reviewed-by: Nikolay Borisov <nik.borisov@suse.com>

Thanks!

> 
> 
> <snip>
> 
>> diff --git a/arch/x86/tools/featuremasks.awk 
>> b/arch/x86/tools/featuremasks.awk
>> new file mode 100755
>> index 000000000000..989b021e73d3
>> --- /dev/null
>> +++ b/arch/x86/tools/featuremasks.awk
>> @@ -0,0 +1,84 @@
>> +#!/usr/bin/awk
>> +#
>> +# Convert cpufeatures.h to a list of compile-time masks
>> +# Note: this blithly assumes that each word has at least one
>> +# feature defined in it; if not, something else is wrong!
>> +#
>> +
>> +BEGIN {
>> +    printf "#ifndef _ASM_X86_FEATUREMASKS_H\n";
>> +    printf "#define _ASM_X86_FEATUREMASKS_H\n\n";
>> +
>> +    file = 0
>> +}
>> +
>> +BEGINFILE {
>> +    switch (++file) {
>> +    case 1:            # cpufeatures.h
>> +        FPAT = "#[ \t]*[a-z]+|[A-Za-z0-9_]+|[^ \t]";
>> +        break;
>> +    case 2:            # .config
>> +        FPAT = "CONFIG_[A-Z0-9_]+|is not set|[yn]";
>> +        break;
>> +    }
>> +}
>> +
> 
> IMO this script could use a bit of high-level comments. Something like:
> 
> 
> # Create a dictionary of sorts, containing all defined feature bits
>> +file == 1 && $1 ~ /^#[ \t]*define$/ && $2 ~ /^X86_FEATURE_/ &&
>> +$3 == "(" && $5 == "*" && $7 == "+" && $9 == ")" {
>> +    nfeat = $4 * $6 + $8;
>> +    feat = $2;
>> +    sub(/^X86_FEATURE_/, "", feat);
>> +    feats[nfeat] = feat;
>> +}
> 
>> +file == 1 && $1 ~ /^#[ \t]*define$/ && $2 == "NCAPINTS" {
>> +    ncapints = strtonum($3);
>> +}
>> +
> 
> # Create a dictionary featstat[REQUIRED|DISABLED, FEATURE_NAME] = on | off
> 
>> +file == 2 && $1 ~ /^CONFIG_X86_[A-Z]*_FEATURE_/ {
>> +    on = ($2 == "y");
>> +    if (split($1, fs, "CONFIG_X86_|_FEATURE_") == 3)
>> +        featstat[fs[2], fs[3]] = on;
>> +}
>> +
>> +END {
>> +    sets[1] = "REQUIRED";
>> +    sets[2] = "DISABLED";
>> +
>> +    for (ns in sets) {
>> +        s = sets[ns];
>> +
>> +        printf "/*\n";
>> +        printf " * %s features:\n", s;
>> +        printf " *\n";
>> +        fstr = "";
>> +        for (i = 0; i < ncapints; i++) {
>> +            mask = 0;
>> +            for (j = 0; j < 32; j++) {
>> +                nfeat = i*32 + j;
>> +                feat = feats[nfeat];
>> +                if (feat) {
>> +                    st = !!featstat[s, feat];
>> +                    if (st) {
>> +                        nfstr = fstr " " feat;
>> +                        if (length(nfstr) > 72) {
>> +                            printf " *   %s\n", fstr;
>> +                            nfstr = " " feat;
>> +                        }
>> +                        fstr = nfstr;
>> +                    }
>> +                    mask += st * (2 ^ j);
> 
> nit: This expression can be changed to mask += (2 ^j) and moved inside 
> the 'if (st)' branch. Essentially only add a bit iff that status of the 
> relevant feature in the kconfig is y, which is signified by the value of 
> 'st' variable.

I will do a polish and send v4.

> 
>> +                }
>> +            }
>> +            masks[i] = mask;
>> +        }
>> +        printf " *   %s\n */\n\n", fstr;
>> +
>> +        for (i = 0; i < ncapints; i++) {
>> +            printf "#define %s_MASK%d\t0x%08x\n", s, i, masks[i];

I should append 'U' here too.

>> +        }
>> +
>> +        printf "#define %s_MASK_CHECK BUILD_BUG_ON_ZERO(NCAPINTS != 
>> %d)\n\n", s, ncapints;
>> +    }
>> +
>> +    printf "#endif /* _ASM_X86_FEATUREMASKS_H */\n";
>> +}
> 


  reply	other threads:[~2024-06-27 16:46 UTC|newest]

Thread overview: 16+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-06-22 17:14 [PATCH v3 0/4] x86/cpufeatures: Automatically generate required and disabled feature masks Xin Li (Intel)
2024-06-22 17:14 ` [PATCH v3 1/4] x86/cpufeatures: Add {required,disabled} feature configs Xin Li (Intel)
2024-06-22 17:14 ` [PATCH v3 2/4] x86/cpufeatures: Generate a feature mask header based on build config Xin Li (Intel)
2024-06-26 11:36   ` Nikolay Borisov
2024-06-27 16:45     ` Xin Li [this message]
2024-06-22 17:14 ` [PATCH v3 3/4] x86/cpufeatures: Remove {disabled,required}-features.h Xin Li (Intel)
2024-06-22 17:14 ` [PATCH v3 4/4] x86/cpufeatures: Use AWK to generate {REQUIRED|DISABLED}_MASK_BIT_SET Xin Li (Intel)
2024-06-23 20:28   ` Brian Gerst
2024-06-24  7:29     ` Xin Li
2024-06-24 10:24       ` Andrew Cooper
2024-06-24 17:42         ` Xin Li
2024-06-25  5:22         ` [PATCH v3B " Xin Li (Intel)
2024-06-27 18:39           ` Brian Gerst
2024-06-24  9:32     ` [PATCH v3A " Xin Li (Intel)
2024-06-27 16:51     ` [PATCH v3 " Xin Li
2024-06-22 17:31 ` [PATCH v3 0/4] x86/cpufeatures: Automatically generate required and disabled feature masks Xin Li

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=e84aa8f0-3885-452c-9250-c5910d869263@zytor.com \
    --to=xin@zytor.com \
    --cc=acme@kernel.org \
    --cc=akpm@linux-foundation.org \
    --cc=bp@alien8.de \
    --cc=brgerst@gmail.com \
    --cc=dave.hansen@linux.intel.com \
    --cc=hpa@zytor.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-perf-users@vger.kernel.org \
    --cc=mingo@redhat.com \
    --cc=namhyung@kernel.org \
    --cc=nik.borisov@suse.com \
    --cc=peterz@infradead.org \
    --cc=tglx@linutronix.de \
    --cc=will@kernel.org \
    --cc=x86@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®