| Message ID | 20231029183600.451694-5-mirsad.todorovac@alu.unizg.hr (mailing list archive) |
|---|---|
| State | Changes Requested |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | [v5,1/7] r8169: Add r8169_mac_ocp_(write|modify)_seq helpers to reduce spinlock contention | expand |
On 29.10.2023 19:36, Mirsad Goran Todorovac wrote: > Repeated calls to r8168_mac_ocp_write() and r8168_mac_ocp_modify() > in the startup of RTL8168ep involve implicit spin_lock_irqsave() and > spin_unlock_irqrestore() on each invocation. > > This is avoided by grouping unlocked __r8168_mac_ocp_write() and > __r8168_mac_ocp_modify() primitives enclosed in the explicit > spin_lock_irqsave()/spin_unlock_irqrestore() pair. > > Even if the lock is not contended, the check requires a LOCK CMPXCHG > or a similar instruction, which prevents all cores from accessing the > memory bus. > > Fixes: 1287723aa139b ("r8169: add support for RTL8117") > Cc: Heiner Kallweit <hkallweit1@gmail.com> > Cc: Marco Elver <elver@google.com> > Cc: nic_swsd@realtek.com > Cc: "David S. Miller" <davem@davemloft.net> > Cc: Eric Dumazet <edumazet@google.com> > Cc: Jakub Kicinski <kuba@kernel.org> > Cc: Paolo Abeni <pabeni@redhat.com> > Cc: netdev@vger.kernel.org > Cc: linux-kernel@vger.kernel.org > Link: https://lore.kernel.org/lkml/20231028005153.2180411-1-mirsad.todorovac@alu.unizg.hr/ > Link: https://lore.kernel.org/lkml/20231028110459.2644926-1-mirsad.todorovac@alu.unizg.hr/ > Signed-off-by: Mirsad Goran Todorovac <mirsad.todorovac@alu.unizg.hr> > --- > v5: > added unlocked primitives to allow mac ocs modify grouping > applied coalescing of mac ocp writes/modifies for 8168ep and 8117 > some formatting fixes to please checkpatch.pl > > v4: > fixed complaints as advised by Heiner and checkpatch.pl > split the patch into five sections to be more easily manipulated and reviewed > introduced r8168_mac_ocp_write_seq() > applied coalescing of mac ocp writes/modifies for 8168H, 8125 and 8125B > > v3: > removed register/mask pair array sentinels, so using ARRAY_SIZE(). > avoided duplication of RTL_W32() call code as advised by Heiner. > > drivers/net/ethernet/realtek/r8169_main.c | 24 ++++++++++++++--------- > 1 file changed, 15 insertions(+), 9 deletions(-) > > diff --git a/drivers/net/ethernet/realtek/r8169_main.c b/drivers/net/ethernet/realtek/r8169_main.c > index 27016aaeb6a0..50fbacb05953 100644 > --- a/drivers/net/ethernet/realtek/r8169_main.c > +++ b/drivers/net/ethernet/realtek/r8169_main.c > @@ -3348,7 +3348,15 @@ static void rtl_hw_start_8117(struct rtl8169_private *tp) > { 0x19, 0x0040, 0x1100 }, > { 0x59, 0x0040, 0x1100 }, > }; > + > + static const struct e_info_regdata e_info_8117_wr_1[] = { > + { 0xe63e, 0x0001 }, > + { 0xe63e, 0x0000 }, > + { 0xc094, 0x0000 }, > + { 0xc09e, 0x0000 }, > + }; > int rg_saw_cnt; > + unsigned long flags; > > rtl8168ep_stop_cmac(tp); > rtl_ephy_init(tp, e_info_8117); > @@ -3388,15 +3396,13 @@ static void rtl_hw_start_8117(struct rtl8169_private *tp) > r8168_mac_ocp_modify(tp, 0xd412, 0x0fff, sw_cnt_1ms_ini); > } > > - r8168_mac_ocp_modify(tp, 0xe056, 0x00f0, 0x0070); > - r8168_mac_ocp_write(tp, 0xea80, 0x0003); > - r8168_mac_ocp_modify(tp, 0xe052, 0x0000, 0x0009); > - r8168_mac_ocp_modify(tp, 0xd420, 0x0fff, 0x047f); > - > - r8168_mac_ocp_write(tp, 0xe63e, 0x0001); > - r8168_mac_ocp_write(tp, 0xe63e, 0x0000); > - r8168_mac_ocp_write(tp, 0xc094, 0x0000); > - r8168_mac_ocp_write(tp, 0xc09e, 0x0000); > + raw_spin_lock_irqsave(&tp->mac_ocp_lock, flags); > + __r8168_mac_ocp_modify(tp, 0xe056, 0x00f0, 0x0070); > + __r8168_mac_ocp_write(tp, 0xea80, 0x0003); > + __r8168_mac_ocp_modify(tp, 0xe052, 0x0000, 0x0009); > + __r8168_mac_ocp_modify(tp, 0xd420, 0x0fff, 0x047f); > + __r8168_mac_ocp_write_seq(tp, e_info_8117_wr_1); > + raw_spin_unlock_irqrestore(&tp->mac_ocp_lock, flags); > > /* firmware is for MAC only */ > r8169_apply_firmware(tp); Same
diff --git a/drivers/net/ethernet/realtek/r8169_main.c b/drivers/net/ethernet/realtek/r8169_main.c index 27016aaeb6a0..50fbacb05953 100644 --- a/drivers/net/ethernet/realtek/r8169_main.c +++ b/drivers/net/ethernet/realtek/r8169_main.c @@ -3348,7 +3348,15 @@ static void rtl_hw_start_8117(struct rtl8169_private *tp) { 0x19, 0x0040, 0x1100 }, { 0x59, 0x0040, 0x1100 }, }; + + static const struct e_info_regdata e_info_8117_wr_1[] = { + { 0xe63e, 0x0001 }, + { 0xe63e, 0x0000 }, + { 0xc094, 0x0000 }, + { 0xc09e, 0x0000 }, + }; int rg_saw_cnt; + unsigned long flags; rtl8168ep_stop_cmac(tp); rtl_ephy_init(tp, e_info_8117); @@ -3388,15 +3396,13 @@ static void rtl_hw_start_8117(struct rtl8169_private *tp) r8168_mac_ocp_modify(tp, 0xd412, 0x0fff, sw_cnt_1ms_ini); } - r8168_mac_ocp_modify(tp, 0xe056, 0x00f0, 0x0070); - r8168_mac_ocp_write(tp, 0xea80, 0x0003); - r8168_mac_ocp_modify(tp, 0xe052, 0x0000, 0x0009); - r8168_mac_ocp_modify(tp, 0xd420, 0x0fff, 0x047f); - - r8168_mac_ocp_write(tp, 0xe63e, 0x0001); - r8168_mac_ocp_write(tp, 0xe63e, 0x0000); - r8168_mac_ocp_write(tp, 0xc094, 0x0000); - r8168_mac_ocp_write(tp, 0xc09e, 0x0000); + raw_spin_lock_irqsave(&tp->mac_ocp_lock, flags); + __r8168_mac_ocp_modify(tp, 0xe056, 0x00f0, 0x0070); + __r8168_mac_ocp_write(tp, 0xea80, 0x0003); + __r8168_mac_ocp_modify(tp, 0xe052, 0x0000, 0x0009); + __r8168_mac_ocp_modify(tp, 0xd420, 0x0fff, 0x047f); + __r8168_mac_ocp_write_seq(tp, e_info_8117_wr_1); + raw_spin_unlock_irqrestore(&tp->mac_ocp_lock, flags); /* firmware is for MAC only */ r8169_apply_firmware(tp);
Repeated calls to r8168_mac_ocp_write() and r8168_mac_ocp_modify() in the startup of RTL8168ep involve implicit spin_lock_irqsave() and spin_unlock_irqrestore() on each invocation. This is avoided by grouping unlocked __r8168_mac_ocp_write() and __r8168_mac_ocp_modify() primitives enclosed in the explicit spin_lock_irqsave()/spin_unlock_irqrestore() pair. Even if the lock is not contended, the check requires a LOCK CMPXCHG or a similar instruction, which prevents all cores from accessing the memory bus. Fixes: 1287723aa139b ("r8169: add support for RTL8117") Cc: Heiner Kallweit <hkallweit1@gmail.com> Cc: Marco Elver <elver@google.com> Cc: nic_swsd@realtek.com Cc: "David S. Miller" <davem@davemloft.net> Cc: Eric Dumazet <edumazet@google.com> Cc: Jakub Kicinski <kuba@kernel.org> Cc: Paolo Abeni <pabeni@redhat.com> Cc: netdev@vger.kernel.org Cc: linux-kernel@vger.kernel.org Link: https://lore.kernel.org/lkml/20231028005153.2180411-1-mirsad.todorovac@alu.unizg.hr/ Link: https://lore.kernel.org/lkml/20231028110459.2644926-1-mirsad.todorovac@alu.unizg.hr/ Signed-off-by: Mirsad Goran Todorovac <mirsad.todorovac@alu.unizg.hr> --- v5: added unlocked primitives to allow mac ocs modify grouping applied coalescing of mac ocp writes/modifies for 8168ep and 8117 some formatting fixes to please checkpatch.pl v4: fixed complaints as advised by Heiner and checkpatch.pl split the patch into five sections to be more easily manipulated and reviewed introduced r8168_mac_ocp_write_seq() applied coalescing of mac ocp writes/modifies for 8168H, 8125 and 8125B v3: removed register/mask pair array sentinels, so using ARRAY_SIZE(). avoided duplication of RTL_W32() call code as advised by Heiner. drivers/net/ethernet/realtek/r8169_main.c | 24 ++++++++++++++--------- 1 file changed, 15 insertions(+), 9 deletions(-)