From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mail-wr1-f49.google.com (mail-wr1-f49.google.com [209.85.221.49]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 42A4839C65C for ; Wed, 4 Feb 2026 10:17:09 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.221.49 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1770200229; cv=none; b=lyLbDISmog+J8PpAvz/IFo++n04EV3tfs4/0gPCl2KexXyB3ZwjB81umLS/Yk8Rz13St5KqCnF0y8vvbC8i8ZIjVkwOD6eQ929NkR8SKRvNkeR3Hq3/ZK9VrAl7xZN1nGr6jGGkYd8mErtF9OHsbDkJf+2ljhS7Lf+DiCROL1zQ= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1770200229; c=relaxed/simple; bh=AzFKCNCSVtzlEDccPu0VHE8JIaqRNCIYrx/kv3SaMLc=; h=Date:From:To:Cc:Subject:Message-ID:In-Reply-To:References: MIME-Version:Content-Type; b=AN9Hwk4GYLA6E2jRx2Z8hwI5TmX+X6Im7wU9hEG28koBlgdySv/GLihYxhBzx624PlrrjrPncWsjQzGruBKyco0lqGrwnmi3wGsBKlyN8auu3n3Veco9i6oLmiCQYwgalvNcV8Qx8x+Z8MhIW1a8nmvcGmavLaXJ6gBhnhg4PNY= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=gmail.com; spf=pass smtp.mailfrom=gmail.com; dkim=pass (2048-bit key) header.d=gmail.com header.i=@gmail.com header.b=F8vI9+b/; arc=none smtp.client-ip=209.85.221.49 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=gmail.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=gmail.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=gmail.com header.i=@gmail.com header.b="F8vI9+b/" Received: by mail-wr1-f49.google.com with SMTP id ffacd0b85a97d-4359228b7c6so4662242f8f.2 for ; Wed, 04 Feb 2026 02:17:09 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20230601; t=1770200228; x=1770805028; darn=vger.kernel.org; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:subject:cc:to:from:date:from:to:cc:subject:date :message-id:reply-to; bh=c7U/6fxo9sUKrIYVs11WWmpcleKYiSKhA/fh2M/QTSM=; b=F8vI9+b/Vumisso0QRJ2/Go4h7OPiijz3hIkOCwQx9mgapPvQNKNqe0IlRcJPyGFAO hosVV+0q92kCA9UjzpE+pQAGMwRMKFqaJtWZtJmcvdWc3V41Adry9Z+AtpwjIkY9J7tJ LiKcWnCl4tXK/MvP5vHTEDTMzkwNqeaEtlxTEAIeHlHBarH2wlOf+DikTO1snAIFsiRq /6a0S0wW6EIrEIye/nQngrR68vawDjQ2Pf0WxiXGxN9G9hwbSRUmU3jk5HbDj0nL8v4b XJGkq07FTqTChhDy62Z4PDiyb1Bm7vRjipmPN9IRXmiZkxyNcN9eKRuAU59cA8/aAf5k gvWg== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1770200228; x=1770805028; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:subject:cc:to:from:date:x-gm-gg:x-gm-message-state:from :to:cc:subject:date:message-id:reply-to; bh=c7U/6fxo9sUKrIYVs11WWmpcleKYiSKhA/fh2M/QTSM=; b=ubudWI8vsMCDOpGKeZzttMHwBGgedHC3bQkzRWx4U4mjupwaNxKFEO/nJqTvB1i6WG Pmcm3BnsZKcyPgtofhOvxr1/s2o7j1FckgRrTOf0qrFGNM19vi+8Oe4LYpkYZMR42YJX FqOinJKfe8f7xOFnxiUhzz1mZTZCd1Qe7zruDwyB8JI5nrnQyKQZYIsbgPDCCF+QaHQh AyVEQdgatjApNcv2DLTGukhX6OtFfiSOni298YtR9q+GAOkLE67EZBEN3Gkn4fsZxs6g Pvm/urQ/YtHoNWqvETcnoUmKm4mp+OuFtVfExxEeRp25zlEGc4t9dmol6jiQOdK/grX/ xM7w== X-Forwarded-Encrypted: i=1; AJvYcCWhy7G0tu/kKP3tdVbKLRB6umqyVIYMiZazthkKZQ48bUKykSzxR+WvPgF5Zx+f2QhaOPOCN+i/5CVL0QE=@vger.kernel.org X-Gm-Message-State: AOJu0YwFzmiMHgLCzsvxtvJTri5urTeaYCdClfpWZOr7wqx3VtmJI1eo SNhtWEyjsNK3Nu1drh0Pl01nAZm8396NmdeFYUCZPrXf7GtMd859iKHSd4YLFA== X-Gm-Gg: AZuq6aJOZX5LA6fjKF2Lj9JVXqqIzUhpHlx9p3UuabqhOCGY/gSa6tqmb3MTwu5Uri1 KK2Ox6VWvP89DbX1rQinUCjxY5zGm2maJokK0mtzeFJN3Sdv2cjvVW/XS5v1B9SoJwthL3gpJ7c d9wzWnL2bIjud3oBi3/aFcyB5fWPclBZFSkZOCWgRbitfULGp7yjHQ+5nS70asbq7EnhfHC/jiM H0xRR8AQFOXT+MP1nWIfCZbnsD5z3Rx43KiRhMk0Rvw/e8hqnqdng5yTVU9zcNSBBwJHFfppkun X4ayPkz/k6ZFZmSfL8tSFl/BBc4YzzL1jloE37LypbQ9H234ISP5+dPB3Ip9ULHLSFy4hUrh3T9 xWUys9Q8olsF1VFgvYChPxGPkzekeS6yrGhyhwNRQr89WXxZB8q0y93EUBRuCPyxdYXPcTQBEki xqBObf05/W2mJHkzmfdiQpXMx6rZTVZBpN0kv5zspx6J2Pf43cL6MQ X-Received: by 2002:a5d:5d84:0:b0:435:9d3f:92d7 with SMTP id ffacd0b85a97d-43617e353b1mr3102192f8f.2.1770200227322; Wed, 04 Feb 2026 02:17:07 -0800 (PST) Received: from pumpkin (82-69-66-36.dsl.in-addr.zen.co.uk. [82.69.66.36]) by smtp.gmail.com with ESMTPSA id ffacd0b85a97d-43617e37e78sm5684917f8f.10.2026.02.04.02.17.06 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Wed, 04 Feb 2026 02:17:06 -0800 (PST) Date: Wed, 4 Feb 2026 10:17:05 +0000 From: David Laight To: Willy Tarreau Cc: Thomas =?UTF-8?B?V2Vpw59zY2h1aA==?= , linux-kernel@vger.kernel.org, Cheng Li Subject: Re: [PATCH next 06/12] tools/nolibc/printf: Add support for left alignment and %[tzLq]d" Message-ID: <20260204101705.11d2d99b@pumpkin> In-Reply-To: References: <20260203103000.20206-1-david.laight.linux@gmail.com> <20260203103000.20206-7-david.laight.linux@gmail.com> X-Mailer: Claws Mail 4.1.1 (GTK 3.24.38; arm-unknown-linux-gnueabihf) Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit On Wed, 4 Feb 2026 05:14:51 +0100 Willy Tarreau wrote: > Hi David, > > On Tue, Feb 03, 2026 at 10:29:54AM +0000, david.laight.linux@gmail.com wrote: > > From: David Laight > > > > Use a single 'flags' variable to hold both format flags and length modifiers. > > Use (1u << (c & 31)) for the flag bits to reduce code complexity. > > > > Add support for left justifying fields. > > > > Add support for length modifiers 't' and 'z' (both long) and 'q' and 'L' > > (both long long). > > > > Unconditionall generate the signed values (for %d) to remove a second > > set of checks for the size. > > > > Use 'signed int' for the lengths to make the pad test simpler. > > > > Signed-off-by: David Laight > > --- > > tools/include/nolibc/stdio.h | 88 ++++++++++++++++++++---------------- > > 1 file changed, 50 insertions(+), 38 deletions(-) > > > > diff --git a/tools/include/nolibc/stdio.h b/tools/include/nolibc/stdio.h > > index 164d2384978e..1ce4d357a802 100644 > > --- a/tools/include/nolibc/stdio.h > > +++ b/tools/include/nolibc/stdio.h > > @@ -240,20 +240,20 @@ char *fgets(char *s, int size, FILE *stream) > > } > > > > > > -/* minimal printf(). It supports the following formats: > > - * - %[l*]{d,u,c,x,p} > > - * - %s > > - * - unknown modifiers are ignored. > > +/* simple printf(). It supports the following formats: > > + * - %[-][width][{l,t,z,ll,L,j,q}]{d,u,c,x,p,s,m} > > + * - %% > > + * - invalid formats are copied to the output buffer > > */ > > typedef int (*__nolibc_printf_cb)(void *state, const char *buf, size_t size); > > > > +#define __PF_FLAG(c) (1u << ((c) & 0x1f)) > > This flag will be exposed to user code, you'll have to previx it with > _NOLIBC_. The lines are long enough already, something shorter would be ideal. The #undef at the bottom stops it being exposed. > > > static __attribute__((unused, format(printf, 3, 0))) > > int __nolibc_printf(__nolibc_printf_cb cb, void *state, const char *fmt, va_list args) > > { > > - char lpref, c; > > - unsigned long long v; > > - unsigned int written, width; > > - size_t len; > > + char c; > > + int len, written, width; > > + unsigned int flags; > > char tmpbuf[21]; > > const char *outstr; > > > > @@ -265,6 +265,7 @@ int __nolibc_printf(__nolibc_printf_cb cb, void *state, const char *fmt, va_list > > break; > > > > width = 0; > > + flags = 0; > > if (c != '%') { > > while (*fmt && *fmt != '%') > > fmt++; > > @@ -274,6 +275,13 @@ int __nolibc_printf(__nolibc_printf_cb cb, void *state, const char *fmt, va_list > > > > c = *fmt++; > > > > + /* Flag characters */ > > + for (; c >= 0x20 && c <= 0x3f; c = *fmt++) { > > + if ((__PF_FLAG(c) & (__PF_FLAG('-'))) == 0) > > + break; > > + flags |= __PF_FLAG(c); > > + } > > Honestly I don't find that it improves readability here and makes one > keep doubts in background as "what if c == 'm' which will match '-'?". > I think that one would better be written as the usual: > > for (; c >= 0x20 && c <= 0x3f; c = *fmt++) { > if (c == '-') > break; > flags |= __PF_FLAG(c); > } > > Or even simpler since there's already a condition in the for() loop: > > for (; c >= 0x20 && c != '-' && c <= 0x3f; c = *fmt++) > flags |= __PF_FLAG(c); It is all written that way for when more flags get added - look at the later patches which check for any of "#-+ 0". At that point the line get long and unreadable - as below :-) I didn't want to add the flags here before supporting them later. But they could all be accepted and ignored until implemented. That might be better anyway. Actually it might be worth s/c/ch/ to make the brain see the difference between c and 'c' more easily. Perhaps I'm expand the comment a bit. It is all a hint as to what is happening later on with the character tests. > > > /* width */ > > while (c >= '0' && c <= '9') { > > width *= 10; > > @@ -282,41 +290,34 @@ int __nolibc_printf(__nolibc_printf_cb cb, void *state, const char *fmt, va_list > > c = *fmt++; > > } > > > > - /* Length modifiers */ > > - if (c == 'l') { > > - lpref = 1; > > - c = *fmt++; > > - if (c == 'l') { > > - lpref = 2; > > + /* Length modifiers are lower case except 'L' which is the same a 'q' */ > > + if ((c >= 'a' && c <= 'z') || (c == 'L' && (c = 'q'))) { > > Then please say "... which we replace by 'q'" so that we don't first read > that (c = 'q') as a possible typo. Also maybe add a mention to the fact > that "flags" will exclusively represent lowercase modifiers from now on? I'll clarify it. > > > + if (__PF_FLAG(c) & (__PF_FLAG('l') | __PF_FLAG('t') | __PF_FLAG('z') | > > + __PF_FLAG('j') | __PF_FLAG('q'))) { > > Even though I understand the value in checking bit positions (I use that > all the time as well), above this is just unreadable. Maybe you need a > different macro, maybe define another macro _NOLIBC_PF_LEN_MOD made of > the addition of all the flags to test against, I don't know, but the > construct, the line break in the middle of the expression and the > parenthesis needed for the macro just requires a lot of effort to > understand what's being tested. Alternately another possibility would > be to have another macro taking 4-5 char args and composing the flags > in one call, passing 0 or -1 for unused ones. This would also make > several parenthesis disappear which would help. Hmmm, some macro magic might work, loosely: #define FLNZ(q) (q ? 1 << (q & 31) ? 0) #define FLM3(q1, q2, q3, ...) ((FLNZ(q1) | FLNZ(q2) | FLNZ(q3)) #define FLT(fl, ...) (fl & FLM3(__VA_ARGS__, 0, 0, 0)) #define CT(c, ...) FLT(1 << (c & 31), __VA_ARGS__) Then the above would be: if (CT(c, 'l', 't', 'z', 'j', 'q')) { Clearly needs some better and longer names and a big comment block. > > > + if (c == 'l' && fmt[0] == 'l') { > > + fmt++; > > + c = 'q'; > > + } > > + /* These all miss "# -0+" */ > > + flags |= __PF_FLAG(c); > > c = *fmt++; > > } > > - } else if (c == 'j') { > > - /* intmax_t is long long */ > > - lpref = 2; > > - c = *fmt++; > > - } else { > > - lpref = 0; > > } > > > > if (c == 'c' || c == 'd' || c == 'u' || c == 'x' || c == 'p') { > > + unsigned long long v; > > + long long signed_v; > > char *out = tmpbuf; > > > > - if (c == 'p') > > + if ((c == 'p') || (flags & (__PF_FLAG('l') | __PF_FLAG('t') | __PF_FLAG('z')))) { > > v = va_arg(args, unsigned long); > > - else if (lpref) { > > - if (lpref > 1) > > - v = va_arg(args, unsigned long long); > > - else > > - v = va_arg(args, unsigned long); > > - } else > > + signed_v = (long)v; > > + } else if (flags & (__PF_FLAG('j') | __PF_FLAG('q'))) { > > + v = va_arg(args, unsigned long long); > > + signed_v = v; > > + } else { > > v = va_arg(args, unsigned int); > > - > > - if (c == 'd') { > > - /* sign-extend the value */ > > - if (lpref == 0) > > - v = (long long)(int)v; > > - else if (lpref == 1) > > - v = (long long)(long)v; > > + signed_v = (int)v; > > } > > > > switch (c) { > > @@ -325,7 +326,7 @@ int __nolibc_printf(__nolibc_printf_cb cb, void *state, const char *fmt, va_list > > out[1] = 0; > > break; > > case 'd': > > - i64toa_r(v, out); > > + i64toa_r(signed_v, out); > > break; > > case 'u': > > u64toa_r(v, out); > > @@ -366,14 +367,24 @@ int __nolibc_printf(__nolibc_printf_cb cb, void *state, const char *fmt, va_list > > > > written += len; > > > > - while (width > len) { > > - unsigned int pad_len = ((width - len - 1) & 15) + 1; > > + /* An OPTIMIZER_HIDE_VAR() seems to stop gcc back-merging this > > + * code into one of the conditionals above. > > Be careful, space indentation above. > > > + */ > > + __asm__ volatile("" : "=r"(len) : "0"(len)); > > and here. Oops... > > > + > > + /* Output 'left pad', 'value' then 'right pad'. */ > > + flags &= __PF_FLAG('-'); > > + width -= len; > > + if (flags && cb(state, outstr, len) != 0) > > + return -1; > > + while (width > 0) { > > + int pad_len = ((width - 1) & 15) + 1; > > width -= pad_len; > > written += pad_len; > > if (cb(state, " ", pad_len) != 0) > > return -1; > > } > > - if (cb(state, outstr, len) != 0) > > + if (!flags && cb(state, outstr, len) != 0) > > return -1; > > } > > > > @@ -382,6 +393,7 @@ int __nolibc_printf(__nolibc_printf_cb cb, void *state, const char *fmt, va_list > > > > return written; > > } > > +#undef _PF_FLAG > > This one can be dropped once named as _NOLIBC_xxx I'll see if I can get a max of 2 expansions on a line. Otherwise the lines get horribly long. David > > Willy