From: "Jan Beulich" <JBeulich@suse.com>
To: Daniel Kiper <daniel.kiper@oracle.com>
Cc: Juergen Gross <JGross@suse.com>,
grub-devel@gnu.org, wei.liu2@citrix.com, ian.campbell@citrix.com,
stefano.stabellini@eu.citrix.com, andrew.cooper3@citrix.com,
roy.franz@linaro.org, ning.sun@intel.com,
david.vrabel@citrix.com, phcoder@gmail.com,
xen-devel@lists.xenproject.org, qiaowei.ren@intel.com,
keir@xen.org, richard.l.maliszewski@intel.com,
gang.wei@intel.com, fu.wei@linaro.org
Subject: Re: [PATCH v2 21/23] x86/boot: implement early command line parser in C
Date: Thu, 27 Aug 2015 06:43:39 -0600 [thread overview]
Message-ID: <55DF221B020000780009D6C6__16036.9965983138$1440679525$gmane$org@prv-mh.provo.novell.com> (raw)
In-Reply-To: <1437402558-7313-22-git-send-email-daniel.kiper@oracle.com>
>>> On 20.07.15 at 16:29, <daniel.kiper@oracle.com> wrote:
> Current early command line parser implementation in assembler
> is very difficult to change to relocatable stuff using segment
> registers. This requires a lot of changes in very weird and
> fragile code. So, reimplement this functionality in C. This
> way code will be relocatable out of the box and much easier
> to maintain.
All appreciated and nice, but the goal of making the code
relocatable by playing with segment registers sounds fragile:
This breaks assumptions the compiler may validly make.
> xen/arch/x86/boot/cmdline.S | 367 -------------------------------------
> xen/arch/x86/boot/cmdline.c | 396 ++++++++++++++++++++++++++++++++++++++++
A fundamental expectation I would have had is for the C file to be
noticeably smaller than the assembly file.
> --- /dev/null
> +++ b/xen/arch/x86/boot/cmdline.c
>[...]
> +#define VESA_WIDTH 0
> +#define VESA_HEIGHT 1
> +#define VESA_DEPTH 2
> +
> +#define VESA_SIZE 3
These should go away in favor of using individual (sub)structure fields.
> +#define __cdecl __attribute__((__cdecl__))
???
> +#define __packed __attribute__((__packed__))
> +#define __text __attribute__((__section__(".text")))
> +#define __used __attribute__((__used__))
Likely better to include compiler.h instead.
> +#define max(x,y) ({ \
> + const typeof(x) _x = (x); \
> + const typeof(y) _y = (y); \
> + (void) (&_x == &_y); \
> + _x > _y ? _x : _y; })
I also wonder whether -imacros .../xen/kernel.h wouldn't be a better
approach here. Please really think hard on how to avoid duplications
like these.
> +#define strlen_static(s) (sizeof(s) - 1)
What is this good for? A decent compiler should be able to deal with
strlen("..."). Plus your macro is longer that what it tries to "abbreviate".
> +static const char empty_chars[] __text = " \n\r\t";
What is empty about them? DYM blank or (white) space or separator
or delimiter? I also wonder whether \n and \r are actually usefully here,
as they should (if at all) only end the line.
> +/**
> + * strlen - Find the length of a string
> + * @s: The string to be sized
> + */
> +static size_t strlen(const char *s)
Comments are certainly nice, but in this special case I'd rather suggest
against bloating the code by commenting standard library functions.
> +static int strtoi(const char *s, const char *stop, const char **next)
> +{
> + int base = 10, i, ores = 0, res = 0;
> +
> + if ( *s == '0' )
> + base = (tolower(*++s) == 'x') ? (++s, 16) : 8;
> +
> + for ( ; *s != '\0'; ++s )
> + {
> + for ( i = 0; stop && stop[i] != '\0'; ++i )
> + if ( *s == stop[i] )
> + goto out;
> +
> + if ( *s < '0' || (*s > '7' && base == 8) )
> + {
> + res = -1;
> + goto out;
> + }
> +
> + if ( *s > '9' && (base != 16 || tolower(*s) < 'a' || tolower(*s) > 'f') )
> + {
> + res = -1;
> + goto out;
> + }
> +
> + res *= base;
> + res += (tolower(*s) >= 'a') ? (tolower(*s) - 'a' + 10) : (*s - '0');
> +
> + if ( ores > res )
> + {
> + res = -1;
> + goto out;
> + }
> +
> + ores = res;
> + }
> +
> +out:
C labels intended by at least one space please.
> +static const char *find_opt(const char *cmdline, const char *opt, int arg)
> +{
> + size_t lc, lo;
> + static const char mm[] __text = "--";
I'd be surprised if there weren't compiler/assembler versions
complaining about a section type conflict here. I can see why you
want everything in one section, but I'd rather suggest achieving
this at the linking step (which I would suppose to already be taking
care of this).
> +static u8 skip_realmode(const char *cmdline)
> +{
> + static const char nrm[] __text = "no-real-mode";
> + static const char tboot[] __text = "tboot=";
> +
> + if ( find_opt(cmdline, nrm, 0) || find_opt(cmdline, tboot, 1) )
> + return 1;
> +
> + return 0;
return find_opt(cmdline, nrm, 0) || find_opt(cmdline, tboot, 1);
> +static u8 edd_parse(const char *cmdline)
> +{
> + const char *c;
> + size_t la;
> + static const char edd[] __text = "edd=";
> + static const char edd_off[] __text = "off";
> + static const char edd_skipmbr[] __text = "skipmbr";
> +
> + c = find_opt(cmdline, edd, 1);
> +
> + if ( !c )
> + return 0;
> +
> + c += strlen_static(edd);
> + la = strcspn(c, empty_chars);
> +
> + if ( !strncmp(c, edd_off, max(la, strlen_static(edd_off))) )
> + return 2;
> + else if ( !strncmp(c, edd_skipmbr, max(la, strlen_static(edd_skipmbr))) )
Pointless else.
> + return 1;
> +
> + return 0;
And the last two returns can be folded again anyway.
> +static void __cdecl __used cmdline_parse_early(const char *cmdline, early_boot_opts_t *ebo)
I don't see the point of the __cdecl, and (as said before) dislike the
static __used pair.
> --- a/xen/arch/x86/boot/head.S
> +++ b/xen/arch/x86/boot/head.S
> @@ -429,8 +429,24 @@ trampoline_setup:
> cmp $sym_phys(__trampoline_seg_stop),%edi
> jb 1b
>
> + /* Do not parse command line on EFI platform here. */
> + cmpb $1,sym_phys(skip_realmode)
Is there a reason you can't look at efi.flags instead here (and in
the other case you abuse skip_realmode as meaning EFI)?
> --- a/xen/arch/x86/boot/trampoline.S
> +++ b/xen/arch/x86/boot/trampoline.S
> @@ -1,3 +1,5 @@
> +#include "video.h"
Please move this farther down, making invisible all its definitions to
code not supposed to use them.
Jan
next prev parent reply other threads:[~2015-08-27 12:43 UTC|newest]
Thread overview: 119+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <1437402558-7313-1-git-send-email-daniel.kiper@oracle.com>
2015-07-20 14:28 ` [PATCH v2 01/23] x86/boot: remove unneeded instruction Daniel Kiper
2015-07-20 14:28 ` [PATCH v2 02/23] x86/boot: copy only text section from *.lnk file to *.bin file Daniel Kiper
2015-07-20 14:28 ` [PATCH v2 03/23] x86: zero BSS using stosl instead of stosb Daniel Kiper
2015-07-20 14:28 ` [PATCH v2 04/23] x86/boot: call reloc() using cdecl calling convention Daniel Kiper
2015-07-20 14:29 ` [PATCH v2 05/23] x86/boot/reloc: create generic alloc and copy functions Daniel Kiper
2015-07-20 14:29 ` [PATCH v2 06/23] x86/boot: use %ecx instead of %eax Daniel Kiper
2015-07-20 14:29 ` [PATCH v2 07/23] x86/boot/reloc: Rename some variables and rearrange code a bit Daniel Kiper
2015-07-20 14:29 ` [PATCH v2 08/23] x86: add multiboot2 protocol support Daniel Kiper
2015-08-10 19:17 ` Konrad Rzeszutek Wilk
[not found] ` <20150810191747.GE13576@l.oracle.com>
2015-08-13 19:22 ` Daniel Kiper
[not found] ` <20150813192212.GI3503@olila.local.net-space.pl>
2015-08-14 10:03 ` Jan Beulich
[not found] ` <55CDD8F4020000780009AF3C@prv-mh.provo.novell.com>
2015-08-15 6:00 ` Andrew Cooper
2015-08-18 8:12 ` Jan Beulich
2015-08-18 12:00 ` Daniel Kiper
2015-08-18 14:20 ` Jan Beulich
2015-07-20 14:29 ` [PATCH v2 09/23] efi: create efi_enabled() Daniel Kiper
2015-08-10 19:20 ` Konrad Rzeszutek Wilk
2015-08-20 15:18 ` Jan Beulich
2015-08-22 12:33 ` Daniel Kiper
[not found] ` <20150822123359.GX7143@olila.local.net-space.pl>
2015-08-24 11:29 ` Jan Beulich
2015-07-20 14:29 ` [PATCH v2 10/23] efi: build xen.gz with EFI code Daniel Kiper
2015-07-20 14:29 ` [PATCH v2 11/23] efi: split out efi_init() Daniel Kiper
2015-08-10 19:25 ` Konrad Rzeszutek Wilk
2015-07-20 14:29 ` [PATCH v2 12/23] efi: split out efi_console_set_mode() Daniel Kiper
2015-07-20 14:29 ` [PATCH v2 13/23] efi: split out efi_get_gop() Daniel Kiper
2015-07-20 14:29 ` [PATCH v2 14/23] efi: split out efi_find_gop_mode() Daniel Kiper
2015-08-10 19:31 ` Konrad Rzeszutek Wilk
2015-08-20 15:48 ` Jan Beulich
2015-07-20 14:29 ` [PATCH v2 15/23] efi: split out efi_tables() Daniel Kiper
2015-08-10 19:32 ` Konrad Rzeszutek Wilk
2015-07-20 14:29 ` [PATCH v2 16/23] efi: split out efi_variables() Daniel Kiper
2015-08-10 19:34 ` Konrad Rzeszutek Wilk
2015-07-20 14:29 ` [PATCH v2 17/23] efi: split out efi_set_gop_mode() Daniel Kiper
2015-08-10 19:34 ` Konrad Rzeszutek Wilk
2015-07-20 14:29 ` [PATCH v2 18/23] efi: split out efi_exit_boot() Daniel Kiper
2015-07-20 14:29 ` [PATCH v2 19/23] x86/efi: create new early memory allocator Daniel Kiper
2015-07-20 14:29 ` [PATCH v2 20/23] x86: add multiboot2 protocol support for EFI platforms Daniel Kiper
2015-08-10 20:07 ` Konrad Rzeszutek Wilk
2015-08-11 15:23 ` Konrad Rzeszutek Wilk
2015-08-27 12:01 ` Jan Beulich
[not found] ` <55DF1836020000780009D674@prv-mh.provo.novell.com>
2015-09-22 15:21 ` Daniel Kiper
2015-09-22 15:58 ` Jan Beulich
2015-07-20 14:29 ` [PATCH v2 21/23] x86/boot: implement early command line parser in C Daniel Kiper
2015-07-20 14:29 ` [PATCH v2 22/23] x86: make Xen early boot code relocatable Daniel Kiper
2015-07-20 14:29 ` [PATCH v2 23/23] x86: add multiboot2 protocol support for relocatable images Daniel Kiper
2015-08-11 16:56 ` Konrad Rzeszutek Wilk
[not found] ` <20150811165658.GC32231@l.oracle.com>
2015-08-14 11:57 ` Daniel Kiper
[not found] ` <20150814115701.GB8034@olila.local.net-space.pl>
2015-08-14 13:43 ` Konrad Rzeszutek Wilk
[not found] ` <1437402558-7313-3-git-send-email-daniel.kiper@oracle.com>
2015-07-21 9:35 ` [PATCH v2 02/23] x86/boot: copy only text section from *.lnk file to *.bin file Jan Beulich
2015-07-21 17:23 ` Daniel Kiper
2015-07-22 5:14 ` Jan Beulich
2015-07-22 8:02 ` Jan Beulich
2015-07-22 13:31 ` Daniel Kiper
2015-07-22 14:07 ` Jan Beulich
[not found] ` <1437402558-7313-4-git-send-email-daniel.kiper@oracle.com>
2015-07-21 9:37 ` [PATCH v2 03/23] x86: zero BSS using stosl instead of stosb Jan Beulich
2015-07-21 18:23 ` Daniel Kiper
2015-07-22 5:18 ` Jan Beulich
2015-07-22 8:42 ` Andrew Cooper
2015-07-22 10:04 ` Jan Beulich
2015-07-22 11:22 ` Andrew Cooper
2015-07-22 11:48 ` Jan Beulich
2015-07-21 9:39 ` [PATCH v2 00/23] x86: multiboot2 protocol support Jan Beulich
[not found] ` <1437402558-7313-2-git-send-email-daniel.kiper@oracle.com>
2015-07-24 16:22 ` [PATCH v2 01/23] x86/boot: remove unneeded instruction Konrad Rzeszutek Wilk
[not found] ` <20150724162257.GB2220@l.oracle.com>
2015-07-27 19:46 ` Daniel Kiper
[not found] ` <20150727194608.GB3492@olila.local.net-space.pl>
2015-08-10 16:07 ` Konrad Rzeszutek Wilk
[not found] ` <1437402558-7313-5-git-send-email-daniel.kiper@oracle.com>
2015-08-10 16:33 ` [PATCH v2 04/23] x86/boot: call reloc() using cdecl calling convention Konrad Rzeszutek Wilk
2015-08-17 15:44 ` Jan Beulich
[not found] ` <1437402558-7313-7-git-send-email-daniel.kiper@oracle.com>
2015-08-10 16:36 ` [PATCH v2 06/23] x86/boot: use %ecx instead of %eax Konrad Rzeszutek Wilk
[not found] ` <1437402558-7313-8-git-send-email-daniel.kiper@oracle.com>
2015-08-10 16:40 ` [PATCH v2 07/23] x86/boot/reloc: Rename some variables and rearrange code a bit Konrad Rzeszutek Wilk
2015-08-17 15:55 ` Jan Beulich
[not found] ` <1437402558-7313-11-git-send-email-daniel.kiper@oracle.com>
2015-08-10 19:24 ` [PATCH v2 10/23] efi: build xen.gz with EFI code Konrad Rzeszutek Wilk
2015-08-20 15:39 ` Jan Beulich
2015-08-22 13:59 ` Daniel Kiper
[not found] ` <20150822135919.GY7143@olila.local.net-space.pl>
2015-08-24 11:35 ` Jan Beulich
2015-08-24 20:54 ` Daniel Kiper
[not found] ` <20150824205445.GG7143@olila.local.net-space.pl>
2015-08-25 10:50 ` Andrew Cooper
2015-08-25 12:09 ` Jan Beulich
[not found] ` <55DC485D.4030904@citrix.com>
2015-08-25 15:39 ` Daniel Kiper
[not found] ` <55DC7705020000780009CE76@prv-mh.provo.novell.com>
2015-08-25 16:31 ` Daniel Kiper
[not found] ` <20150825163141.GJ7143@olila.local.net-space.pl>
2015-08-26 6:46 ` Jan Beulich
[not found] ` <55DD7CDE020000780009D086@prv-mh.provo.novell.com>
2015-08-26 12:33 ` Daniel Kiper
[not found] ` <20150826123356.GB10861@olila.local.net-space.pl>
2015-08-26 12:40 ` Jan Beulich
[not found] ` <55DDCFD6020000780009D226@prv-mh.provo.novell.com>
2015-08-26 12:58 ` Daniel Kiper
[not found] ` <1437402558-7313-13-git-send-email-daniel.kiper@oracle.com>
2015-08-10 19:25 ` [PATCH v2 12/23] efi: split out efi_console_set_mode() Konrad Rzeszutek Wilk
[not found] ` <1437402558-7313-14-git-send-email-daniel.kiper@oracle.com>
2015-08-10 19:27 ` [PATCH v2 13/23] efi: split out efi_get_gop() Konrad Rzeszutek Wilk
[not found] ` <1437402558-7313-19-git-send-email-daniel.kiper@oracle.com>
2015-08-10 19:36 ` [PATCH v2 18/23] efi: split out efi_exit_boot() Konrad Rzeszutek Wilk
[not found] ` <1437402558-7313-20-git-send-email-daniel.kiper@oracle.com>
2015-08-10 19:49 ` [PATCH v2 19/23] x86/efi: create new early memory allocator Konrad Rzeszutek Wilk
2015-08-27 11:23 ` Jan Beulich
[not found] ` <1437402558-7313-6-git-send-email-daniel.kiper@oracle.com>
2015-08-17 15:51 ` [PATCH v2 05/23] x86/boot/reloc: create generic alloc and copy functions Jan Beulich
[not found] ` <55D21F3E020000780009B716@prv-mh.provo.novell.com>
2015-08-17 22:03 ` Daniel Kiper
[not found] ` <1437402558-7313-22-git-send-email-daniel.kiper@oracle.com>
2015-08-10 20:31 ` [PATCH v2 21/23] x86/boot: implement early command line parser in C Konrad Rzeszutek Wilk
2015-08-11 14:43 ` Konrad Rzeszutek Wilk
2015-08-27 12:43 ` Jan Beulich [this message]
[not found] ` <55DF221B020000780009D6C6@prv-mh.provo.novell.com>
2015-09-22 17:03 ` Daniel Kiper
[not found] ` <20150922170332.GH3501@olila.local.net-space.pl>
2015-09-23 7:25 ` Jan Beulich
[not found] ` <1437402558-7313-23-git-send-email-daniel.kiper@oracle.com>
2015-08-11 16:48 ` [PATCH v2 22/23] x86: make Xen early boot code relocatable Konrad Rzeszutek Wilk
[not found] ` <20150811164806.GB32231@l.oracle.com>
2015-08-14 11:52 ` Daniel Kiper
[not found] ` <20150814115205.GA8034@olila.local.net-space.pl>
2015-08-14 12:49 ` Jan Beulich
[not found] ` <55CDFFEE020000780009B0B0@prv-mh.provo.novell.com>
2015-08-14 13:59 ` Daniel Kiper
[not found] ` <20150814135904.GC8034@olila.local.net-space.pl>
2015-08-14 14:32 ` Jan Beulich
[not found] ` <55CE1805020000780009B1C0@prv-mh.provo.novell.com>
2015-08-14 14:37 ` Daniel Kiper
[not found] ` <20150814143729.GE8034@olila.local.net-space.pl>
2015-08-14 15:12 ` Jan Beulich
2015-08-14 15:20 ` Konrad Rzeszutek Wilk
2015-08-27 13:12 ` Jan Beulich
[not found] ` <55DF28E6020000780009D6E4@prv-mh.provo.novell.com>
2015-08-27 15:10 ` Daniel Kiper
[not found] ` <20150827151054.GI10944@olila.local.net-space.pl>
2015-08-27 15:29 ` Jan Beulich
[not found] ` <55DF48FB020000780009D83F@prv-mh.provo.novell.com>
2015-08-27 17:56 ` Ben Hildred
2015-08-27 18:04 ` Andrew Cooper
[not found] ` <55DF5125.20708@citrix.com>
2015-08-28 6:54 ` Jan Beulich
[not found] ` <55E021C4020000780009DA95@prv-mh.provo.novell.com>
2015-08-28 11:59 ` Andrew Cooper
[not found] ` <CAKcyEGpdnJUXA=uW4cNdcLn9xS7fUDowzzCbhn6c1yTh4WkPGg@mail.gmail.com>
2015-08-28 8:22 ` Jan Beulich
[not found] ` <55E03676020000780009DAFB@prv-mh.provo.novell.com>
2015-08-28 13:42 ` Konrad Rzeszutek Wilk
[not found] ` <20150828134214.GC2412@l.oracle.com>
2015-08-28 14:16 ` Jan Beulich
2015-08-31 19:49 ` Daniel Kiper
[not found] ` <20150831194956.GJ4530@olila.local.net-space.pl>
2015-09-01 6:59 ` Jan Beulich
2015-08-28 14:24 ` Jan Beulich
2015-08-16 13:48 [PATCH v2 21/23] x86/boot: implement early command line parser in C George Diamantopoulos
2015-08-16 17:22 ` Konrad Rzeszutek Wilk
2015-08-17 20:37 ` Daniel Kiper
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='55DF221B020000780009D6C6__16036.9965983138$1440679525$gmane$org@prv-mh.provo.novell.com' \
--to=jbeulich@suse.com \
--cc=JGross@suse.com \
--cc=andrew.cooper3@citrix.com \
--cc=daniel.kiper@oracle.com \
--cc=david.vrabel@citrix.com \
--cc=fu.wei@linaro.org \
--cc=gang.wei@intel.com \
--cc=grub-devel@gnu.org \
--cc=ian.campbell@citrix.com \
--cc=keir@xen.org \
--cc=ning.sun@intel.com \
--cc=phcoder@gmail.com \
--cc=qiaowei.ren@intel.com \
--cc=richard.l.maliszewski@intel.com \
--cc=roy.franz@linaro.org \
--cc=stefano.stabellini@eu.citrix.com \
--cc=wei.liu2@citrix.com \
--cc=xen-devel@lists.xenproject.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;
as well as URLs for NNTP newsgroup(s).