The Linux Kernel Mailing List
 help / color / mirror / Atom feed
From: Linus Torvalds <torvalds@linux-foundation.org>
To: Roland Dreier <rdreier@cisco.com>, Ingo Molnar <mingo@elte.hu>,
	Ian Campbell <ian.campbell@citrix.com>,
	Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Cc: Helge Deller <deller@gmx.de>,
	Rusty Russell <rusty@rustcorp.com.au>,
	linux-parisc <linux-parisc@vger.kernel.org>,
	Linux Kernel Development <linux-kernel@vger.kernel.org>,
	Kyle McMartin <kyle@mcmartin.ca>,
	Randolph Chung <randolph@tausq.org>,
	Andrew Morton <akpm@linux-foundation.org>,
	Sam Ravnborg <sam@ravnborg.org>,
	John David Anglin <dave@hiauly1.hia.nrc.ca>
Subject: Re: [PATCH] parisc: fix module loading failure of large kernel modules (take 4)
Date: Wed, 31 Dec 2008 10:02:49 -0800 (PST)	[thread overview]
Message-ID: <alpine.LFD.2.00.0812310953440.5086@localhost.localdomain> (raw)
In-Reply-To: <alpine.LFD.2.00.0812310942110.5086@localhost.localdomain>



On Wed, 31 Dec 2008, Linus Torvalds wrote:
> 
> I guess I could make a sparse rule for this, but nobody seems to run or 
> care about sparse anyway. Sad.

Ho humm. Looks like a sparse rule might be in order. These are just from 
my (smallish) personal config. I haven't checked them all, but I did check 
the first two, and yes, sparse did get it right.

  init/main.c:548:24: warning: Calling weak function 'smp_setup_processor_id' in file scope
  init/main.c:674:24: warning: Calling weak function 'thread_info_cache_init' in file scope
  arch/x86/kernel/e820.c:1364:38: warning: Calling weak function 'machine_specific_memory_setup' in file scope
  arch/x86/kernel/e820.c:1371:20: warning: Calling weak function 'memory_setup' in file scope
  arch/x86/kernel/reboot.c:386:22: warning: Calling weak function 'mach_reboot_fixups' in file scope
  kernel/sched.c:7720:26: warning: Calling weak function 'arch_update_cpu_topology' in file scope
  kernel/sched.c:7808:41: warning: Calling weak function 'arch_update_cpu_topology' in file scope
  kernel/fork.c:229:29: warning: Calling weak function 'arch_dup_task_struct' in file scope
  kernel/fork.c:1325:44: warning: Calling weak function 'idle_regs' in file scope
  kernel/sched_clock.c:158:19: warning: Calling weak function 'sched_clock' in file scope
  kernel/sched_clock.c:207:19: warning: Calling weak function 'sched_clock' in file scope
  kernel/time/timekeeping.c:283:43: warning: Calling weak function 'read_persistent_clock' in file scope
  kernel/time/timekeeping.c:318:43: warning: Calling weak function 'read_persistent_clock' in file scope
  kernel/time/timekeeping.c:353:50: warning: Calling weak function 'read_persistent_clock' in file scope
  mm/vmalloc.c:1645:18: warning: Calling weak function 'vmalloc_sync_all' in file scope
  mm/hugetlb.c:1039:32: warning: Calling weak function 'alloc_bootmem_huge_page' in file scope
  mm/sparse.c:448:29: warning: Calling weak function 'vmemmap_populate_print_last' in file scope
  fs/pipe.c:1069:18: warning: Calling weak function 'sys_pipe2' in file scope
  fs/proc/meminfo.c:145:21: warning: Calling weak function 'arch_report_meminfo' in file scope
  drivers/acpi/pci_irq.c:676:21: warning: Calling weak function 'acpi_unregister_gsi' in file scope
  drivers/char/mem.c:168:23: warning: Calling weak function 'unxlate_dev_mem_ptr' in file scope
  drivers/char/mem.c:172:22: warning: Calling weak function 'unxlate_dev_mem_ptr' in file scope
  drivers/char/mem.c:240:23: warning: Calling weak function 'unxlate_dev_mem_ptr' in file scope
  drivers/char/mem.c:246:22: warning: Calling weak function 'unxlate_dev_mem_ptr' in file scope
  drivers/char/mem.c:318:12: warning: Calling weak function 'map_devmem' in file scope
  drivers/char/mem.c:324:14: warning: Calling weak function 'unmap_devmem' in file scope
  drivers/char/mem.c:349:35: warning: Calling weak function 'phys_mem_access_prot_allowed' in file scope
  drivers/char/mem.c:365:15: warning: Calling weak function 'unmap_devmem' in file scope
  drivers/pci/probe.c:1159:36: warning: Calling weak function 'set_pci_bus_resources_arch_default' in file scope
  drivers/pci/pci.c:1015:24: warning: Calling weak function 'pcibios_disable_device' in file scope
  drivers/pci/pci.c:1043:37: warning: Calling weak function 'pcibios_set_pcie_reset_state' in file scope
  drivers/pci/pci-sysfs.c:897:39: warning: Calling weak function 'pcibios_add_platform_entries' in file scope
  drivers/pci/msi.c:49:27: warning: Calling weak function 'arch_setup_msi_irq' in file scope
  drivers/pci/msi.c:69:25: warning: Calling weak function 'arch_teardown_msi_irq' in file scope
  drivers/pci/msi.c:421:27: warning: Calling weak function 'arch_setup_msi_irqs' in file scope
  drivers/pci/msi.c:492:27: warning: Calling weak function 'arch_setup_msi_irqs' in file scope
  drivers/pci/msi.c:562:29: warning: Calling weak function 'arch_msi_check_device' in file scope
  drivers/pci/msi.c:654:24: warning: Calling weak function 'arch_teardown_msi_irqs' in file scope
  lib/swiotlb.c:141:28: warning: Calling weak function 'swiotlb_phys_to_bus' in file scope
  lib/swiotlb.c:146:41: warning: Calling weak function 'swiotlb_bus_to_phys' in file scope
  lib/swiotlb.c:156:28: warning: Calling weak function 'swiotlb_phys_to_bus' in file scope
  lib/swiotlb.c:167:30: warning: Calling weak function 'swiotlb_phys_to_bus' in file scope
  lib/swiotlb.c:168:28: warning: Calling weak function 'swiotlb_phys_to_bus' in file scope
  lib/swiotlb.c:204:35: warning: Calling weak function 'swiotlb_alloc_boot' in file scope
  lib/swiotlb.c:260:31: warning: Calling weak function 'swiotlb_alloc' in file scope
  lib/swiotlb.c:336:58: warning: Calling weak function 'swiotlb_arch_range_needs_mapping' in file scope
  lib/swiotlb.c:336:58: warning: Calling weak function 'swiotlb_arch_range_needs_mapping' in file scope

I bet there are others that I don't see just because I don't compile the 
code.

The appended trial sparse diff is against sparse -git as of a couple of 
days ago if anybody wants to run it themselves.

The _logical_ fix would be to add "noinline" to the definition of 
"__weak", but that's reported not to help. I think the bug may be in the 
assembler that pre-links the call if it sees it in file scope. Or maybe 
gcc doesn't honor noinline for empty functions. Whatever.

Regardless, it's a damn pain.

			Linus

---
 evaluate.c |    7 +++++++
 parse.c    |    7 ++++---
 symbol.h   |    3 ++-
 3 files changed, 13 insertions(+), 4 deletions(-)

diff --git a/evaluate.c b/evaluate.c
index f976645..0ae6c93 100644
--- a/evaluate.c
+++ b/evaluate.c
@@ -2744,6 +2744,13 @@ static int evaluate_symbol_call(struct expression *expr)
 	if (ctype->op && ctype->op->evaluate)
 		return ctype->op->evaluate(expr);
 
+	if (ctype->ctype.modifiers & MOD_WEAK) {
+		struct symbol *fn = ctype->ctype.base_type;
+
+		if (fn->stmt || fn->inline_stmt)
+			warning(expr->pos, "Calling weak function '%s' in file scope", show_ident(ctype->ident));
+	}
+
 	if (ctype->ctype.modifiers & MOD_INLINE) {
 		int ret;
 		struct symbol *curr = current_fn;
diff --git a/parse.c b/parse.c
index eb31871..1ea497f 100644
--- a/parse.c
+++ b/parse.c
@@ -289,6 +289,9 @@ static struct init_keyword {
 	{ "word",	NS_KEYWORD,	MOD_LONG,	.op = &mode_spec_op },
 	{ "__word__",	NS_KEYWORD,	MOD_LONG,	.op = &mode_spec_op },
 
+	{ "weak",	NS_KEYWORD,	MOD_WEAK,	.op = &attr_mod_op },
+	{ "__weak__",	NS_KEYWORD,	MOD_WEAK,	.op = &attr_mod_op },
+
 	/* Ignored attributes */
 	{ "nothrow",	NS_KEYWORD,	.op = &ignore_attr_op },
 	{ "__nothrow",	NS_KEYWORD,	.op = &ignore_attr_op },
@@ -317,8 +320,6 @@ static struct init_keyword {
 	{ "__sentinel__",	NS_KEYWORD,	.op = &ignore_attr_op },
 	{ "regparm",	NS_KEYWORD,	.op = &ignore_attr_op },
 	{ "__regparm__",	NS_KEYWORD,	.op = &ignore_attr_op },
-	{ "weak",	NS_KEYWORD,	.op = &ignore_attr_op },
-	{ "__weak__",	NS_KEYWORD,	.op = &ignore_attr_op },
 	{ "alias",	NS_KEYWORD,	.op = &ignore_attr_op },
 	{ "__alias__",	NS_KEYWORD,	.op = &ignore_attr_op },
 	{ "pure",	NS_KEYWORD,	.op = &ignore_attr_op },
@@ -1582,7 +1583,7 @@ static struct statement *start_function(struct symbol *sym)
 	start_function_scope();
 	ret = alloc_symbol(sym->pos, SYM_NODE);
 	ret->ctype = sym->ctype.base_type->ctype;
-	ret->ctype.modifiers &= ~(MOD_STORAGE | MOD_CONST | MOD_VOLATILE | MOD_INLINE | MOD_ADDRESSABLE | MOD_NOCAST | MOD_NODEREF | MOD_ACCESSED | MOD_TOPLEVEL);
+	ret->ctype.modifiers &= ~(MOD_STORAGE | MOD_CONST | MOD_VOLATILE | MOD_INLINE | MOD_ADDRESSABLE | MOD_NOCAST | MOD_NODEREF | MOD_ACCESSED | MOD_TOPLEVEL | MOD_WEAK);
 	ret->ctype.modifiers |= (MOD_AUTO | MOD_REGISTER);
 	bind_symbol(ret, &return_ident, NS_ITERATOR);
 	stmt->ret = ret;
diff --git a/symbol.h b/symbol.h
index c4d7f28..b3fcccd 100644
--- a/symbol.h
+++ b/symbol.h
@@ -185,6 +185,7 @@ struct symbol {
 #define MOD_LONGLONG	0x0800
 
 #define MOD_TYPEDEF	0x1000
+#define MOD_WEAK	0x2000
 
 #define MOD_INLINE	0x40000
 #define MOD_ADDRESSABLE	0x80000
@@ -205,7 +206,7 @@ struct symbol {
 #define MOD_BITWISE	0x80000000
 
 #define MOD_NONLOCAL	(MOD_EXTERN | MOD_TOPLEVEL)
-#define MOD_STORAGE	(MOD_AUTO | MOD_REGISTER | MOD_STATIC | MOD_EXTERN | MOD_INLINE | MOD_TOPLEVEL | MOD_FORCE)
+#define MOD_STORAGE	(MOD_AUTO | MOD_REGISTER | MOD_STATIC | MOD_EXTERN | MOD_INLINE | MOD_TOPLEVEL | MOD_FORCE | MOD_WEAK)
 #define MOD_SIGNEDNESS	(MOD_SIGNED | MOD_UNSIGNED | MOD_EXPLICITLY_SIGNED)
 #define MOD_SPECIFIER	(MOD_CHAR | MOD_SHORT | MOD_LONG | MOD_LONGLONG | MOD_SIGNEDNESS)
 #define MOD_SIZE	(MOD_CHAR | MOD_SHORT | MOD_LONG | MOD_LONGLONG)

  reply	other threads:[~2008-12-31 18:03 UTC|newest]

Thread overview: 60+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-12-29 20:34 [PATCH] parisc: fix module loading failure of large kernel modules (take 2) Helge Deller
2008-12-29 20:43 ` [PATCH 1/2] module.c: fix module loading failure of large " Helge Deller
     [not found]   ` <20081230180724.GA15235@bombadil.infradead.org>
2008-12-30 18:10     ` Kyle McMartin
2008-12-30 18:18       ` Helge Deller
2008-12-30 19:42         ` [PATCH 1/2] module.c: fix module loading failure of large modules (take 3) Helge Deller
2008-12-29 20:45 ` [PATCH 2/2] parisc: fix module loading failure of large modules (take 2) Helge Deller
2008-12-30 19:55   ` [PATCH 2/2] parisc: fix module loading failure of large modules (take 3) Helge Deller
2008-12-30 22:45 ` [PATCH] parisc: fix module loading failure of large kernel modules (take 2) Rusty Russell
2008-12-30 23:02   ` Helge Deller
2008-12-31  4:08     ` Rusty Russell
2008-12-31 11:31   ` [PATCH] parisc: fix module loading failure of large kernel modules (take 4) Helge Deller
2008-12-31 11:36     ` [PATCH 2/2] parisc: fix module loading failure of large modules Helge Deller
2008-12-31 13:32     ` [PATCH] parisc: fix module loading failure of large kernel modules (take 4) Rusty Russell
2008-12-31 14:13       ` Helge Deller
2009-01-01  0:52         ` Rusty Russell
2009-01-01 12:02           ` Helge Deller
2008-12-31 17:29     ` Linus Torvalds
2008-12-31 17:36       ` Roland Dreier
2008-12-31 17:47         ` Linus Torvalds
2008-12-31 18:02           ` Linus Torvalds [this message]
2008-12-31 18:11           ` Sam Ravnborg
2009-01-02 11:31             ` Ingo Molnar
2008-12-31 18:54           ` Andrew Morton
2008-12-31 21:22             ` Linus Torvalds
2008-12-31 22:14               ` David Miller
2009-01-02 11:55                 ` [PATCH] kbuild: Disallow GCC 4.1.0 / 4.1.1 Ingo Molnar
2009-01-02 13:43                   ` Bartlomiej Zolnierkiewicz
2009-01-02 15:21                     ` [PATCH] kbuild: Remove gcc 4.1.0 quirk from init/main.c Ingo Molnar
2009-01-02 18:05                       ` Sam Ravnborg
2009-01-02 16:49                   ` [PATCH] kbuild: Disallow GCC 4.1.0 / 4.1.1 Linus Torvalds
2009-01-02 17:38                     ` Linus Torvalds
2009-01-02 17:46                       ` Ingo Molnar
2009-01-02 17:54                         ` [PATCH] Disallow gcc versions 3.{0,1} Ingo Molnar
2009-01-02 17:58                         ` [PATCH] kbuild: Disallow GCC 4.1.0 / 4.1.1 Linus Torvalds
2009-01-02 18:01                           ` Ingo Molnar
2009-01-02 18:05                             ` Linus Torvalds
2009-01-02 18:08                               ` Linus Torvalds
2009-01-02 19:54                               ` Willy Tarreau
2009-01-02 20:18                                 ` Linus Torvalds
2009-01-02 17:57                       ` Sam Ravnborg
2009-01-02 18:04                         ` Linus Torvalds
2009-01-02 18:27                           ` Sam Ravnborg
2009-01-02 18:28                             ` Randy Dunlap
2009-01-02 18:51                           ` Al Viro
2009-01-02 19:14                             ` Andi Kleen
2009-01-02 22:52                               ` Al Viro
2009-01-03 14:03                           ` Krzysztof Halasa
2009-01-02 18:22                         ` Ingo Molnar
2009-01-02 18:29                           ` Sam Ravnborg
2009-01-02 18:33                             ` Ingo Molnar
2009-01-02 19:05                               ` Detlef Riekenberg
2009-01-02 22:27                         ` Benjamin Herrenschmidt
2009-01-02 22:37                           ` Sam Ravnborg
2009-01-02 17:44                     ` Ingo Molnar
2009-01-01 14:24               ` [PATCH] parisc: fix module loading failure of large kernel modules (take 4) Ingo Molnar
2009-01-01 16:37                 ` Andrew Morton
2008-12-31 17:39       ` Helge Deller
2008-12-31 18:24       ` James Bottomley
2008-12-31 22:16       ` Rusty Russell
2009-01-01  7:12       ` Jeremy Fitzhardinge

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=alpine.LFD.2.00.0812310953440.5086@localhost.localdomain \
    --to=torvalds@linux-foundation.org \
    --cc=akpm@linux-foundation.org \
    --cc=dave@hiauly1.hia.nrc.ca \
    --cc=deller@gmx.de \
    --cc=ian.campbell@citrix.com \
    --cc=jeremy.fitzhardinge@citrix.com \
    --cc=kyle@mcmartin.ca \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-parisc@vger.kernel.org \
    --cc=mingo@elte.hu \
    --cc=randolph@tausq.org \
    --cc=rdreier@cisco.com \
    --cc=rusty@rustcorp.com.au \
    --cc=sam@ravnborg.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