| 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: <Pine.LNX.4.44L0.0611171428590.2627-100000@iolanthe.rowland.org>
Date: Fri, 17 Nov 2006 14:35:46 -0500 (EST)
From: Alan Stern <stern@...land.harvard.edu>
To: Daniel Walker <dwalker@...sta.com>, Andrew Morton <akpm@...l.org>
cc: Venkatesh Pallipadi <venkatesh.pallipadi@...el.com>,
Jens Axboe <axboe@...nel.dk>,
Christoph Lameter <clameter@....com>,
"David S. Miller" <davem@...emloft.net>,
"Paul E. McKenney" <paulmck@...ibm.com>,
Kernel development list <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] Allow NULL pointers in percpu_free
The patch (as824b) makes percpu_free() ignore NULL arguments, as one
would expect for a deallocation routine. (Note that free_percpu is
#defined as percpu_free in include/linux/percpu.h.) A few callers are
updated to remove now-unneeded tests for NULL. A few other callers
already seem to assume that passing a NULL pointer to percpu_free() is
okay!
The patch also removes an unnecessary NULL check in percpu_depopulate().
Signed-off-by: Alan Stern <stern@...land.harvard.edu>
---
On Fri, 17 Nov 2006, Daniel Walker wrote:
> > > Should be unlikely() right?
> >
> > It certainly could be. I tend not to put such annotations in my code, but
> > it wouldn't hurt.
>
> It's actually a really good idea to add them .. I've noticed they tend
> to make my kernels smaller, although I wouldn't expect that to always be
> the case..
Smaller???!!! How on earth do you think that could happen? When you
don't use unlikely() there are fewer restrictions on the compiler; hence
it should be free to generate smaller code without the annotation than
with it. Unless maybe you don't set CONFIG_CC_OPTIMIZE_FOR_SIZE, which
ought to be a much better way of getting small kernels...
> Another reason is that in -mm we can track how often this
> condition is triggered with likely profiling. With kfree, for instance,
> there were a number of callers that frequently called kfree(NULL), which
> IMO isn't good.
Okay, here's the revised patch.
Alan Stern
Index: usb-2.6/arch/i386/kernel/acpi/cstate.c
===================================================================
--- usb-2.6.orig/arch/i386/kernel/acpi/cstate.c
+++ usb-2.6/arch/i386/kernel/acpi/cstate.c
@@ -155,10 +155,8 @@ static int __init ffh_cstate_init(void)
static void __exit ffh_cstate_exit(void)
{
- if (cpu_cstate_entry) {
- free_percpu(cpu_cstate_entry);
- cpu_cstate_entry = NULL;
- }
+ free_percpu(cpu_cstate_entry);
+ cpu_cstate_entry = NULL;
}
arch_initcall(ffh_cstate_init);
Index: usb-2.6/block/blktrace.c
===================================================================
--- usb-2.6.orig/block/blktrace.c
+++ usb-2.6/block/blktrace.c
@@ -366,8 +366,7 @@ err:
if (bt) {
if (bt->dropped_file)
debugfs_remove(bt->dropped_file);
- if (bt->sequence)
- free_percpu(bt->sequence);
+ free_percpu(bt->sequence);
if (bt->rchan)
relay_close(bt->rchan);
kfree(bt);
Index: usb-2.6/mm/allocpercpu.c
===================================================================
--- usb-2.6.orig/mm/allocpercpu.c
+++ usb-2.6/mm/allocpercpu.c
@@ -17,10 +17,9 @@
void percpu_depopulate(void *__pdata, int cpu)
{
struct percpu_data *pdata = __percpu_disguise(__pdata);
- if (pdata->ptrs[cpu]) {
- kfree(pdata->ptrs[cpu]);
- pdata->ptrs[cpu] = NULL;
- }
+
+ kfree(pdata->ptrs[cpu]);
+ pdata->ptrs[cpu] = NULL;
}
EXPORT_SYMBOL_GPL(percpu_depopulate);
@@ -123,6 +122,8 @@ EXPORT_SYMBOL_GPL(__percpu_alloc_mask);
*/
void percpu_free(void *__pdata)
{
+ if (unlikely(!__pdata))
+ return;
__percpu_depopulate_mask(__pdata, &cpu_possible_map);
kfree(__percpu_disguise(__pdata));
}
Index: usb-2.6/net/ipv6/af_inet6.c
===================================================================
--- usb-2.6.orig/net/ipv6/af_inet6.c
+++ usb-2.6/net/ipv6/af_inet6.c
@@ -719,10 +719,8 @@ snmp6_mib_free(void *ptr[2])
{
if (ptr == NULL)
return;
- if (ptr[0])
- free_percpu(ptr[0]);
- if (ptr[1])
- free_percpu(ptr[1]);
+ free_percpu(ptr[0]);
+ free_percpu(ptr[1]);
ptr[0] = ptr[1] = NULL;
}
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists