Message ID | 20241028-open-coded-timeouts-v2-1-c7294bb845a1@linux.microsoft.com |
---|---|
State | New |
Headers | show |
Series | Converge on secs_to_jiffies() | expand |
On 10/29/2024 10:25 AM, Thomas Gleixner wrote: > On Tue, Oct 29 2024 at 17:22, Geert Uytterhoeven wrote: >> On Tue, Oct 29, 2024 at 5:08 PM Thomas Gleixner <tglx@linutronix.de> wrote: >>> On Mon, Oct 28 2024 at 19:11, Easwar Hariharan wrote: >>>> diff --git a/include/linux/jiffies.h b/include/linux/jiffies.h >>>> index 1220f0fbe5bf..e5256bb5f851 100644 >>>> --- a/include/linux/jiffies.h >>>> +++ b/include/linux/jiffies.h >>>> @@ -526,6 +526,8 @@ static __always_inline unsigned long msecs_to_jiffies(const unsigned int m) >>>> } >>>> } >>>> >>>> +#define secs_to_jiffies(_secs) ((_secs) * HZ) >>> >>> Can you please make that a static inline, as there is no need for macro >>> magic like in the other conversions, and add a kernel doc comment which >>> documents this? >> >> Note that a static inline means it cannot be used in e.g. struct initializers, >> which are substantial users of "<value> * HZ". > > Bah. That wants to be mentioned in the change log then. > > Still the macro should be documented. > > Thanks, > > tglx Thanks for the review, I'll add a kernel doc in v3 and mention the limitations of an inline function. - Easwar
From: Thomas Gleixner > Sent: 29 October 2024 17:25 > > On Tue, Oct 29 2024 at 17:22, Geert Uytterhoeven wrote: > > On Tue, Oct 29, 2024 at 5:08 PM Thomas Gleixner <tglx@linutronix.de> wrote: > >> On Mon, Oct 28 2024 at 19:11, Easwar Hariharan wrote: > >> > diff --git a/include/linux/jiffies.h b/include/linux/jiffies.h > >> > index 1220f0fbe5bf..e5256bb5f851 100644 > >> > --- a/include/linux/jiffies.h > >> > +++ b/include/linux/jiffies.h > >> > @@ -526,6 +526,8 @@ static __always_inline unsigned long msecs_to_jiffies(const unsigned int m) > >> > } > >> > } > >> > > >> > +#define secs_to_jiffies(_secs) ((_secs) * HZ) > >> > >> Can you please make that a static inline, as there is no need for macro > >> magic like in the other conversions, and add a kernel doc comment which > >> documents this? > > > > Note that a static inline means it cannot be used in e.g. struct initializers, > > which are substantial users of "<value> * HZ". > > Bah. That wants to be mentioned in the change log then. > > Still the macro should be documented. I was wondering if it really had any purpose at all. It just obfuscates code, doesn't even make it smaller. David - Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)
On Wed, Nov 06 2024 at 22:19, David Laight wrote: > From: Thomas Gleixner >> Still the macro should be documented. > > I was wondering if it really had any purpose at all. > It just obfuscates code, doesn't even make it smaller. What is obfuscated here? secs_to_jiffies() is clearly describing what this is about, while 5 * HZ is not. Nothing in a driver has to care about the underlying conversion of a time delta expressed in SI units to a jiffies based time delta. Thanks, tglx
diff --git a/include/linux/jiffies.h b/include/linux/jiffies.h index 1220f0fbe5bf..e5256bb5f851 100644 --- a/include/linux/jiffies.h +++ b/include/linux/jiffies.h @@ -526,6 +526,8 @@ static __always_inline unsigned long msecs_to_jiffies(const unsigned int m) } } +#define secs_to_jiffies(_secs) ((_secs) * HZ) + extern unsigned long __usecs_to_jiffies(const unsigned int u); #if !(USEC_PER_SEC % HZ) static inline unsigned long _usecs_to_jiffies(const unsigned int u) diff --git a/net/bluetooth/hci_event.c b/net/bluetooth/hci_event.c index 0bbad90ddd6f..7b35c58bbbeb 100644 --- a/net/bluetooth/hci_event.c +++ b/net/bluetooth/hci_event.c @@ -42,8 +42,6 @@ #define ZERO_KEY "\x00\x00\x00\x00\x00\x00\x00\x00" \ "\x00\x00\x00\x00\x00\x00\x00\x00" -#define secs_to_jiffies(_secs) msecs_to_jiffies((_secs) * 1000) - /* Handle HCI Event packets */ static void *hci_ev_skb_pull(struct hci_dev *hdev, struct sk_buff *skb,
secs_to_jiffies() is defined in hci_event.c and cannot be reused by other call sites. Hoist it into the core code to allow conversion of the ~1150 usages of msecs_to_jiffies() that either: - use a multiplier value of 1000 or equivalently MSEC_PER_SEC, or - have timeouts that are denominated in seconds (i.e. end in 000) This will also allow conversion of yet more sites that use (sec * HZ) directly, and improve their readability. TO: "K. Y. Srinivasan" <kys@microsoft.com> TO: Haiyang Zhang <haiyangz@microsoft.com> TO: Wei Liu <wei.liu@kernel.org> TO: Dexuan Cui <decui@microsoft.com> TO: linux-hyperv@vger.kernel.org TO: Anna-Maria Behnsen <anna-maria@linutronix.de> TO: Thomas Gleixner <tglx@linutronix.de> TO: Geert Uytterhoeven <geert@linux-m68k.org> TO: Marcel Holtmann <marcel@holtmann.org> TO: Johan Hedberg <johan.hedberg@gmail.com> TO: Luiz Augusto von Dentz <luiz.dentz@gmail.com> TO: linux-bluetooth@vger.kernel.org TO: linux-kernel@vger.kernel.org Suggested-by: Michael Kelley <mhklinux@outlook.com> Signed-off-by: Easwar Hariharan <eahariha@linux.microsoft.com> --- include/linux/jiffies.h | 2 ++ net/bluetooth/hci_event.c | 2 -- 2 files changed, 2 insertions(+), 2 deletions(-)