| Message ID | 1540366553-18541-1-git-send-email-clabbe@baylibre.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | include: add setbits32/clrbits32/clrsetbits32/setbits64/clrbits64/clrsetbits64 | expand |
On Wed, Oct 24, 2018 at 07:35:46AM +0000, Corentin Labbe wrote: > This patchset adds a new set of functions which are open-coded in lot of > place. > Basicly the pattern is always the same, "read, modify a bit, write" > some driver and the powerpc arch already have thoses pattern them as functions. (like ahci_sunxi.c or dwmac-meson8b) The advantage of them being open-coded is that it's _obvious_ to the reviewer that there is a read-modify-write going on which, in a multi- threaded environment, may need some locking (so it should trigger a review of the locking around that code.) With it hidden inside a helper which has no locking itself, it becomes much easier to pass over in review, which means that races are much more likely to go unspotted - and that is bad news.
On Wed, Oct 24, 2018 at 09:57:00AM +0100, Russell King - ARM Linux wrote: > On Wed, Oct 24, 2018 at 07:35:46AM +0000, Corentin Labbe wrote: > > This patchset adds a new set of functions which are open-coded in lot of > > place. > > Basicly the pattern is always the same, "read, modify a bit, write" > > some driver and the powerpc arch already have thoses pattern them as functions. (like ahci_sunxi.c or dwmac-meson8b) > > The advantage of them being open-coded is that it's _obvious_ to the > reviewer that there is a read-modify-write going on which, in a multi- > threaded environment, may need some locking (so it should trigger a > review of the locking around that code.) > > With it hidden inside a helper which has no locking itself, it becomes > much easier to pass over in review, which means that races are much > more likely to go unspotted - and that is bad news. > Hello I understand your fear, but I think the benefit overhaul thoses. Furthermore, drivers which I have converted does not need such locking. If you want I can rename the header to linux/setbits-non-atomic.h for making obvious the lack of locking. Regards
On Thu, Nov 15, 2018 at 10:30:34AM +0100, LABBE Corentin wrote: > On Wed, Oct 24, 2018 at 09:57:00AM +0100, Russell King - ARM Linux wrote: > > On Wed, Oct 24, 2018 at 07:35:46AM +0000, Corentin Labbe wrote: > > > This patchset adds a new set of functions which are open-coded in lot of > > > place. > > > Basicly the pattern is always the same, "read, modify a bit, write" > > > some driver and the powerpc arch already have thoses pattern them as functions. (like ahci_sunxi.c or dwmac-meson8b) > > > > The advantage of them being open-coded is that it's _obvious_ to the > > reviewer that there is a read-modify-write going on which, in a multi- > > threaded environment, may need some locking (so it should trigger a > > review of the locking around that code.) > > > > With it hidden inside a helper which has no locking itself, it becomes > > much easier to pass over in review, which means that races are much > > more likely to go unspotted - and that is bad news. > > > > Hello > > I understand your fear, but I think the benefit overhaul thoses. > Furthermore, drivers which I have converted does not need such locking. > > If you want I can rename the header to linux/setbits-non-atomic.h for making obvious the lack of locking. It'd probably be better in the function name - it then doesn't get "lost" that it's non-atomic when it's included via other headers.
On Thu, Nov 15, 2018 at 09:33:48AM +0000, Russell King - ARM Linux wrote: > On Thu, Nov 15, 2018 at 10:30:34AM +0100, LABBE Corentin wrote: > > On Wed, Oct 24, 2018 at 09:57:00AM +0100, Russell King - ARM Linux wrote: > > > On Wed, Oct 24, 2018 at 07:35:46AM +0000, Corentin Labbe wrote: > > > > This patchset adds a new set of functions which are open-coded in lot of > > > > place. > > > > Basicly the pattern is always the same, "read, modify a bit, write" > > > > some driver and the powerpc arch already have thoses pattern them as functions. (like ahci_sunxi.c or dwmac-meson8b) > > > > > > The advantage of them being open-coded is that it's _obvious_ to the > > > reviewer that there is a read-modify-write going on which, in a multi- > > > threaded environment, may need some locking (so it should trigger a > > > review of the locking around that code.) > > > > > > With it hidden inside a helper which has no locking itself, it becomes > > > much easier to pass over in review, which means that races are much > > > more likely to go unspotted - and that is bad news. > > > > > > > Hello > > > > I understand your fear, but I think the benefit overhaul thoses. > > Furthermore, drivers which I have converted does not need such locking. > > > > If you want I can rename the header to linux/setbits-non-atomic.h for making obvious the lack of locking. > > It'd probably be better in the function name - it then doesn't get > "lost" that it's non-atomic when it's included via other headers. > I proposed that way for doing like writeq have do it with io-64-nonatomic-hi-lo.h
Hello This patchset adds a new set of functions which are open-coded in lot of place. Basicly the pattern is always the same, "read, modify a bit, write" some driver and the powerpc arch already have thoses pattern them as functions. (like ahci_sunxi.c or dwmac-meson8b) The first patch rename some powerpc functions for being consistent with the new name convention. The second patch adds the header with all setbits functions. The third patch is a try to implement a coccinelle semantic patch to find all place where xxxbits function could be used. It should not be merged since it is un-finalized. For the moment, the "add setbits.h header" is not working and need a future coccinelle version. The four last patch are example of some drivers conversion. Thoses patchs give an example of the reduction of code won by using xxxbits32. I would like to thanks Julia Lawall for her help on the coccinelle patch. Note that I dont know which maintainer will take the linux/setbits.h include patch. Regards Changes since v2: - Fixed patch title - Fixed style problems - shorted macro arguments name Changes since v1: - renamed LE functions to _leXX - updated coccinnelle patch with JLawall's comments Corentin Labbe (7): powerpc: rename setbits32/clrbits32 to setbits_be32/clrbits_be32 include: add setbits_leXX/clrbits_leXX/clrsetbits_leXX in linux/setbits.h coccinelle: add xxxsetbits_leXX converting spatch ata: ahci_sunxi: use xxxsetbitsi_le32 functions net: ethernet: stmmac: dwmac-sun8i: use xxxsetbits_le32 drm: meson: use xxxsetbits_le32 net: stmmac: dwmac-meson8b: use xxxsetbits_le32 arch/powerpc/include/asm/fsl_lbc.h | 2 +- arch/powerpc/include/asm/io.h | 4 +- arch/powerpc/platforms/44x/canyonlands.c | 4 +- arch/powerpc/platforms/4xx/gpio.c | 28 +- arch/powerpc/platforms/512x/pdm360ng.c | 6 +- arch/powerpc/platforms/52xx/mpc52xx_common.c | 6 +- arch/powerpc/platforms/52xx/mpc52xx_gpt.c | 12 +- arch/powerpc/platforms/82xx/ep8248e.c | 2 +- arch/powerpc/platforms/82xx/km82xx.c | 6 +- arch/powerpc/platforms/82xx/mpc8272_ads.c | 10 +- arch/powerpc/platforms/82xx/pq2.c | 2 +- arch/powerpc/platforms/82xx/pq2ads-pci-pic.c | 4 +- arch/powerpc/platforms/82xx/pq2fads.c | 10 +- arch/powerpc/platforms/83xx/km83xx.c | 6 +- arch/powerpc/platforms/83xx/mpc836x_mds.c | 2 +- arch/powerpc/platforms/85xx/mpc85xx_mds.c | 2 +- arch/powerpc/platforms/85xx/mpc85xx_pm_ops.c | 4 +- arch/powerpc/platforms/85xx/mpc85xx_rdb.c | 2 +- arch/powerpc/platforms/85xx/p1022_ds.c | 6 +- arch/powerpc/platforms/85xx/p1022_rdk.c | 6 +- arch/powerpc/platforms/85xx/t1042rdb_diu.c | 6 +- arch/powerpc/platforms/85xx/twr_p102x.c | 2 +- arch/powerpc/platforms/86xx/mpc8610_hpcd.c | 6 +- arch/powerpc/platforms/8xx/adder875.c | 2 +- arch/powerpc/platforms/8xx/m8xx_setup.c | 4 +- arch/powerpc/platforms/8xx/mpc86xads_setup.c | 4 +- arch/powerpc/platforms/8xx/mpc885ads_setup.c | 28 +- .../platforms/embedded6xx/flipper-pic.c | 6 +- arch/powerpc/platforms/embedded6xx/hlwd-pic.c | 8 +- arch/powerpc/platforms/embedded6xx/wii.c | 12 +- arch/powerpc/sysdev/cpm1.c | 26 +- arch/powerpc/sysdev/cpm2.c | 16 +- arch/powerpc/sysdev/cpm_common.c | 4 +- arch/powerpc/sysdev/fsl_85xx_l2ctlr.c | 16 +- arch/powerpc/sysdev/fsl_lbc.c | 2 +- arch/powerpc/sysdev/fsl_pci.c | 12 +- arch/powerpc/sysdev/fsl_pmc.c | 2 +- arch/powerpc/sysdev/fsl_rcpm.c | 74 +-- arch/powerpc/sysdev/fsl_rio.c | 4 +- arch/powerpc/sysdev/fsl_rmu.c | 9 +- arch/powerpc/sysdev/mpic_timer.c | 12 +- drivers/ata/ahci_sunxi.c | 62 +-- drivers/gpu/drm/meson/meson_crtc.c | 14 +- drivers/gpu/drm/meson/meson_dw_hdmi.c | 33 +- drivers/gpu/drm/meson/meson_plane.c | 13 +- drivers/gpu/drm/meson/meson_registers.h | 3 - drivers/gpu/drm/meson/meson_venc.c | 13 +- drivers/gpu/drm/meson/meson_venc_cvbs.c | 4 +- drivers/gpu/drm/meson/meson_viu.c | 65 +-- drivers/gpu/drm/meson/meson_vpp.c | 22 +- .../ethernet/stmicro/stmmac/dwmac-meson8b.c | 56 +- .../net/ethernet/stmicro/stmmac/dwmac-sun8i.c | 62 +-- include/linux/setbits.h | 84 +++ scripts/add_new_include_in_source.py | 61 +++ scripts/coccinelle/misc/setbits32.cocci | 487 ++++++++++++++++++ .../coccinelle/misc/setbits32_relaxed.cocci | 487 ++++++++++++++++++ scripts/coccinelle/misc/setbits64.cocci | 487 ++++++++++++++++++ scripts/coccinelle/misc/setbits_dev.cocci | 235 +++++++++ 58 files changed, 2172 insertions(+), 395 deletions(-) create mode 100644 include/linux/setbits.h create mode 100755 scripts/add_new_include_in_source.py create mode 100644 scripts/coccinelle/misc/setbits32.cocci create mode 100644 scripts/coccinelle/misc/setbits32_relaxed.cocci create mode 100644 scripts/coccinelle/misc/setbits64.cocci create mode 100644 scripts/coccinelle/misc/setbits_dev.cocci