| 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
| ||
|
Message-ID: <7e1d99fd.3806.18a40450dee.Coremail.guosy@buaa.edu.cn>
Date: Tue, 29 Aug 2023 15:49:41 +0800 (GMT+08:00)
From: 郭思远 <guosy@...a.edu.cn>
To: "Jesse T" <mr.bossman075@...il.com>
Cc: "Masahiro Yamada" <masahiroy@...nel.org>, sunying@...iscas.ac.cn,
linux-kbuild@...r.kernel.org, linux-kernel@...r.kernel.org,
"Sergey Senozhatsky" <senozhatsky@...omium.org>,
"Randy Dunlap" <rdunlap@...radead.org>, pengpeng@...as.ac.cn
Subject: Re: Re: [PATCH] kconfig: add dependency warning print about invalid
values in verbose mode
I apologize for this, but I cannot reproduce this issue in my environment.
I have tried removing CONFIG_SMP or changing it to n in the
arch/x86/configs/x86_64_defconfig file,and none of them resulted in segfault.
I am using the latest Linux kernel version and compilation toolchain.
In order to better locate and solve this issue, could you provide more information？
> On Sun, Aug 20, 2023 at 10:40AM Jesse T <mr.bossman075@...il.com>wrote:
>
> On Sat, Aug 19, 2023 at 8:59 PM Masahiro Yamada <masahiroy@...nel.org> wrote:
> >
> > On Wed, Aug 9, 2023 at 9:32 AM <sunying@...iscas.ac.cn> wrote:
> > >
> > > From: Ying Sun <sunying@...iscas.ac.cn>
> > >
> > > Add warning about the configuration option's invalid value in verbose mode,
> > > including error causes, mismatch dependency, old and new values,
> > > to help users correct them.
>
> It should also warn about invalid options which were being talked
> about in other places.
> https://lore.kernel.org/all/20230817012007.131868-1-senozhatsky@chromium.org/T/#u
> https://lore.kernel.org/lkml/CAMuHMdWF0LseSGK6=aodXaoK9D16mxok4DDRs=gKoGox8k6zjg@mail.gmail.com/T/#mb5b73532166014ce0a66b0e7e11dbe06528d863c
>
> I can also cause a segfault by removing options such as removing
> `CONFIG_SMP=y` from x86_64_defconfig
>
> > >
> > > Detailed error messages are printed only when the environment variable
> > > is set like "KCONFIG_VERBOSE=1".
> > > By default, the current behavior is not changed.
> > >
> > > Signed-off-by: Siyuan Guo <zy21df106@...a.edu.cn>
> > > Signed-off-by: Ying Sun <sunying@...iscas.ac.cn>
> >
> >
> > This patch is much bigger than I had expected.
> >
> > I did not go through all the lines, but
> > I think the behavior should be improved.
> >
> >
> >
> > [1] A downgrade from 'y' to 'm' should be warned.
> >
> >
> > You only check (sym->dir_dep.tri == no && sym->def[S_DEF_USER].tri != no)
> >
> > but (sym->dir_dep.tri == mod && sym->def[S_DEF_USER].tri == yes)
> > will cause =y to be downgraded to =m.
> > This must be warned.
> >
> >
> >
> >
> > [2] A new CONFIG option should not be warned.
> >
> > $ make defconfig
> >
> > Add the following two lines to a Kconfig file.
> >
> > config THIS_IS_A_NEW_OPTION
> > def_bool y
> >
> >
> > $ make KCONFIG_VERBOSE=1 oldconfig
> >
> > ERROR : THIS_IS_A_NEW_OPTION[n => y] value is invalid
> > due to it has default value
> > #
> > # configuration written to .config
> > #
> >
> >
> >
> >
> > This is totally harmless.
> > I do not understand why it is warned.
> >
> >
> >
> >
> >
> > BTW, your patch subject says
> > "add dependency warning", but your actually code prints "ERROR".
> >
> > Also, check the coding style.
> >
> >
> >
> >
> >
> >
> >
> > > ---
> > > scripts/kconfig/confdata.c | 121 +++++++++++++++++++++++++++++++++++--
> > > scripts/kconfig/lkc.h | 3 +
> > > scripts/kconfig/symbol.c | 82 +++++++++++++++++++++++--
> > > 3 files changed, 195 insertions(+), 11 deletions(-)
> > >
> > > diff --git a/scripts/kconfig/confdata.c b/scripts/kconfig/confdata.c
> > > index 992575f1e976..fa2ae6f63352 100644
> > > --- a/scripts/kconfig/confdata.c
> > > +++ b/scripts/kconfig/confdata.c
> > > @@ -154,6 +154,7 @@ static void conf_message(const char *fmt, ...)
> > >
> > > static const char *conf_filename;
> > > static int conf_lineno, conf_warnings;
> > > +const char *verbose;
> > >
> > > static void conf_warning(const char *fmt, ...)
> > > {
> > > @@ -226,7 +227,7 @@ static const char *conf_get_rustccfg_name(void)
> > > static int conf_set_sym_val(struct symbol *sym, int def, int def_flags, char *p)
> > > {
> > > char *p2;
> > > -
> > > + static const char * const type[] = {"unknown", "bool", "tristate", "int", "hex", "string"};
>
> Add a new line after the declaration.
> Not sure why checkpatch didn't get this.
>
> > > switch (sym->type) {
> > > case S_TRISTATE:
> > > if (p[0] == 'm') {
> > > @@ -246,9 +247,14 @@ static int conf_set_sym_val(struct symbol *sym, int def, int def_flags, char *p)
> > > sym->flags |= def_flags;
> > > break;
> > > }
> > > - if (def != S_DEF_AUTO)
> > > - conf_warning("symbol value '%s' invalid for %s",
> > > + if (def != S_DEF_AUTO) {
> > > + if (verbose)
> > > + conf_warning("symbol value '%s' invalid for %s\n due to its type is %s",
> > > + p, sym->name, type[sym->type]);
>
> normally indented to the opening parenthesis, same with the rest.
>
> > > + else
> > > + conf_warning("symbol value '%s' invalid for %s",
> > > p, sym->name);
> > > + }
> > > return 1;
> > > case S_STRING:
> > > /* No escaping for S_DEF_AUTO (include/config/auto.conf) */
> > > @@ -274,9 +280,14 @@ static int conf_set_sym_val(struct symbol *sym, int def, int def_flags, char *p)
> > > sym->def[def].val = xstrdup(p);
> > > sym->flags |= def_flags;
> > > } else {
> > > - if (def != S_DEF_AUTO)
> > > - conf_warning("symbol value '%s' invalid for %s",
> > > - p, sym->name);
> > > + if (def != S_DEF_AUTO) {
> > > + if (verbose)
> > > + conf_warning("symbol value '%s' invalid for %s\n due to its type is %s",
> > > + p, sym->name, type[sym->type]);
> > > + else
> > > + conf_warning("symbol value '%s' invalid for %s",
> > > + p, sym->name);
> > > + }
> > > return 1;
> > > }
> > > break;
> > > @@ -528,6 +539,7 @@ int conf_read(const char *name)
> > > int conf_unsaved = 0;
> > > int i;
> > >
> > > + verbose = getenv("KCONFIG_VERBOSE");
> > > conf_set_changed(false);
> > >
> > > if (conf_read_simple(name, S_DEF_USER)) {
> > > @@ -559,6 +571,103 @@ int conf_read(const char *name)
> > > continue;
> > > conf_unsaved++;
> > > /* maybe print value in verbose mode... */
> > > + if (verbose) {
> > > + if (sym_is_choice_value(sym)) {
> > > + struct property *prop = sym_get_choice_prop(sym);
> > > + struct symbol *defsym = prop_get_symbol(prop)->curr.val;
> > > +
> > > + if (defsym && defsym != sym) {
> > > + struct gstr gs = str_new();
> > > +
> > > + str_printf(&gs,
> > > + "\nERROR : %s[%c => %c] value is invalid\n",
> > > + sym->name,
> > > + tristate2char[sym->def[S_DEF_USER].tri],
> > > + tristate2char[sym->curr.tri]);
> > > + str_printf(&gs,
> > > + " due to its not the choice default symbol\n");
> > > + str_printf(&gs,
> > > + " the default symbol is %s\n",
> > > + defsym->name);
> > > + fputs(str_get(&gs), stderr);
> > > + }
> > > + } else {
> > > + switch (sym->type) {
> > > + case S_BOOLEAN:
> > > + case S_TRISTATE:
> > > + if (sym->dir_dep.tri == no &&
> > > + sym->def[S_DEF_USER].tri != no) {
> > > + struct gstr gs = str_new();
> > > +
> > > + str_printf(&gs,
> > > + "\nERROR: unmet direct dependencies detected for %s[%c => %c]\n",
> > > + sym->name,
> > > + tristate2char[sym->def[S_DEF_USER].tri],
> > > + tristate2char[sym->curr.tri]);
> > > + str_printf(&gs,
> > > + " Depends on [%c]: ",
> > > + sym->dir_dep.tri == mod ? 'm' : 'n');
> > > + expr_gstr_print(sym->dir_dep.expr, &gs);
> > > + str_printf(&gs, "\n");
> > > + fputs(str_get(&gs), stderr);
> > > + } else if (sym->rev_dep.tri != no) {
> > > + struct gstr gs = str_new();
> > > +
> > > + str_printf(&gs,
> > > + "\nERROR : %s[%c => %c] value is invalid\n",
> > > + sym->name,
> > > + tristate2char[sym->def[S_DEF_USER].tri],
> > > + tristate2char[sym->curr.tri]);
> > > + str_printf(&gs,
> > > + " due to its invisible and it is selected\n");
>
> As Masahiro said this is unnecessary.
> Maybe add a verbosity level for this.
>
> > > + expr_gstr_print_revdep(sym->rev_dep.expr, &gs, yes,
> > > + " Selected by [y]:\n");
> > > + expr_gstr_print_revdep(sym->rev_dep.expr, &gs, mod,
> > > + " Selected by [m]:\n");
> > > + fputs(str_get(&gs), stderr);
> > > + } else {
> > > + sym_validate_default(sym);
> > > + }
> > > + break;
> > > + case S_INT:
> > > + case S_HEX:
> > > + if (sym->dir_dep.tri == no &&
> > > + strcmp((char *)(sym->def[S_DEF_USER].val), "") != 0) {
> > > + struct gstr gs = str_new();
> > > +
> > > + str_printf(&gs,
> > > + "\nERROR: unmet direct dependencies detected for %s\n",
> > > + sym->name);
> > > + str_printf(&gs,
> > > + " Depends on [%c]: ",
> > > + sym->dir_dep.tri == mod ? 'm' : 'n');
> > > + expr_gstr_print(sym->dir_dep.expr, &gs);
> > > + str_printf(&gs, "\n");
> > > + fputs(str_get(&gs), stderr);
> > > + } else {
> > > + sym_validate_default(sym);
> > > + }
> > > + break;
> > > + case S_STRING:
> > > + if (sym->dir_dep.tri == no &&
> > > + strcmp((char *)(sym->def[S_DEF_USER].val), "") != 0) {
> > > + struct gstr gs = str_new();
> > > +
> > > + str_printf(&gs,
> > > + "\nERROR: unmet direct dependencies detected for %s\n",
> > > + sym->name);
> > > + str_printf(&gs,
> > > + " Depends on [%c]: ",
> > > + sym->dir_dep.tri == mod ? 'm' : 'n');
> > > + expr_gstr_print(sym->dir_dep.expr, &gs);
> > > + str_printf(&gs, "\n");
> > > + fputs(str_get(&gs), stderr);
> > > + }
> > > + default:
> > > + break;
> > > + }
> > > + }
> > > + }
> > > }
> > >
> > > for_all_symbols(i, sym) {
> > > diff --git a/scripts/kconfig/lkc.h b/scripts/kconfig/lkc.h
> > > index 471a59acecec..820a47fb4968 100644
> > > --- a/scripts/kconfig/lkc.h
> > > +++ b/scripts/kconfig/lkc.h
> > > @@ -38,6 +38,8 @@ void zconf_initscan(const char *name);
> > > void zconf_nextfile(const char *name);
> > > int zconf_lineno(void);
> > > const char *zconf_curname(void);
> > > +extern const char *verbose;
> > > +static const char tristate2char[3] = {'n', 'm', 'y'};
>
> This seems non-ideal
>
> > >
> > > /* confdata.c */
> > > const char *conf_get_configname(void);
> > > @@ -112,6 +114,7 @@ struct property *sym_get_range_prop(struct symbol *sym);
> > > const char *sym_get_string_default(struct symbol *sym);
> > > struct symbol *sym_check_deps(struct symbol *sym);
> > > struct symbol *prop_get_symbol(struct property *prop);
> > > +void sym_validate_default(struct symbol *sym);
> > >
> > > static inline tristate sym_get_tristate_value(struct symbol *sym)
> > > {
> > > diff --git a/scripts/kconfig/symbol.c b/scripts/kconfig/symbol.c
> > > index 0572330bf8a7..8b11d6ea1d30 100644
> > > --- a/scripts/kconfig/symbol.c
> > > +++ b/scripts/kconfig/symbol.c
> > > @@ -91,6 +91,53 @@ static struct property *sym_get_default_prop(struct symbol *sym)
> > > return NULL;
> > > }
> > >
> > > +void sym_validate_default(struct symbol *sym)
> > > +{
> > > + if (sym->visible == no) {
> > > + struct gstr gs = str_new();
> > > + const char *value = sym_get_string_default(sym);
> > > +
> > > + switch (sym->type) {
> > > + case S_BOOLEAN:
> > > + case S_TRISTATE:
> > > + if (strcmp(value, "n") != 0) {
> > > + str_printf(&gs,
> > > + "\nERROR : %s[%c => %c] value is invalid\n due to it has default value\n",
> > > + sym->name,
> Same here
> > > + tristate2char[sym->def[S_DEF_USER].tri],
> > > + tristate2char[sym->curr.tri]);
> > > + } else if (sym->implied.tri != no) {
> > > + str_printf(&gs,
> > > + "\nERROR : %s[%c => %c] value is invalid\n due to its invisible and has imply value\n",
> > > + sym->name,
> And here
> > > + tristate2char[sym->def[S_DEF_USER].tri],
> > > + tristate2char[sym->curr.tri]);
> > > + str_printf(&gs,
> > > + " Imply : ");
> > > + expr_gstr_print(sym->implied.expr, &gs);
> > > + str_printf(&gs, "\n");
> > > + }
> > > + break;
> > > + case S_STRING:
> > > + case S_INT:
> > > + case S_HEX:
> > > + if (strcmp(value, "") != 0) {
> > > + str_printf(&gs,
> > > + "\nERROR : %s[%s => %s] value is invalid\n",
> > > + sym->name,
> > > + (char *)(sym->def[S_DEF_USER].val),
> > > + (char *)(sym->curr.val));
> > > + str_printf(&gs,
> > > + " due to it has default value\n");
> > > + }
> > > + break;
> > > + default:
> > > + break;
> > > + }
> > > + fputs(str_get(&gs), stderr);
> > > + }
> > > +}
> > > +
> > > struct property *sym_get_range_prop(struct symbol *sym)
> > > {
> > > struct property *prop;
> > > @@ -600,7 +647,8 @@ bool sym_string_valid(struct symbol *sym, const char *str)
> > > bool sym_string_within_range(struct symbol *sym, const char *str)
> > > {
> > > struct property *prop;
> > > - long long val;
> > > + long long val, left, right;
> > > + struct gstr gs = str_new();
> > >
> > > switch (sym->type) {
> > > case S_STRING:
> > > @@ -612,8 +660,20 @@ bool sym_string_within_range(struct symbol *sym, const char *str)
> > > if (!prop)
> > > return true;
> > > val = strtoll(str, NULL, 10);
> > > - return val >= sym_get_range_val(prop->expr->left.sym, 10) &&
> > > - val <= sym_get_range_val(prop->expr->right.sym, 10);
> > > + left = sym_get_range_val(prop->expr->left.sym, 10);
> > > + right = sym_get_range_val(prop->expr->right.sym, 10);
> > > + if (val >= left && val <= right)
> > > + return true;
> > > + if (verbose) {
> > > + str_printf(&gs,
> > > + "\nERROR: unmet range detected for %s\n",
> > > + sym->name);
> > > + str_printf(&gs,
> > > + " symbol value is %lld, the range is (%lld %lld)\n",
> > > + val, left, right);
> > > + fputs(str_get(&gs), stderr);
> > > + }
> > > + return false;
> > > case S_HEX:
> > > if (!sym_string_valid(sym, str))
> > > return false;
> > > @@ -621,8 +681,20 @@ bool sym_string_within_range(struct symbol *sym, const char *str)
> > > if (!prop)
> > > return true;
> > > val = strtoll(str, NULL, 16);
> > > - return val >= sym_get_range_val(prop->expr->left.sym, 16) &&
> > > - val <= sym_get_range_val(prop->expr->right.sym, 16);
> > > + left = sym_get_range_val(prop->expr->left.sym, 16);
> > > + right = sym_get_range_val(prop->expr->right.sym, 16);
> > > + if (val >= left && val <= right)
> > > + return true;
> > > + if (verbose) {
> > > + str_printf(&gs,
> > > + "\nERROR: unmet range detected for %s\n",
> > > + sym->name);
> > > + str_printf(&gs,
> > > + " symbol value is 0x%llx, the range is (0x%llx 0x%llx)\n",
> > > + val, left, right);
> > > + fputs(str_get(&gs), stderr);
> > > + }
> > > + return false;
> > > case S_BOOLEAN:
> > > case S_TRISTATE:
> > > switch (str[0]) {
> > > --
> > > 2.17.1
> > >
> >
> >
> > --
> > Best Regards
> > Masahiro Yamada
Powered by blists - more mailing lists