Skip to contentSkip to conversation

Respond:Patch 0004-Add-support-for-SUNIX-parallel-card

PATCH In review Morris Ku 3 participants 3 messages Mar 29, 2019 – Apr 1, 2019

lore
Share & tools
lore

Morris Ku

author
#0 lore raw

PATCH 1/1 Respond:Patch 0004-Add-support-for-SUNIX-parallel-card

Hi,
Thanks for review, my replies are inline:

---
@@ -0,0 +1,255 @@
+
+On 19.03.19 13:08, Morris Ku wrote:
+
+> diff --git a/mfd/sunix/snx_ieee1284_ops.c b/mfd/sunix/snx_ieee1284_ops.c +> new file mode 100644
+> index 00000000..2dac03fd
+> --- /dev/null
+
+<snip>
+
+> +size_t sunix_parport_ieee1284_read_nibble(struct snx_parport *port, +> +void *buffer, size_t len, int flags)
+> +{
+> + return 0;
+> +}
+> +
+> +size_t sunix_parport_ieee1284_read_byte(struct snx_parport *port, +> +void *buffer, size_t len, int flags)
+> +{
+> + return 0;
+> +}
+> +
+> +size_t sunix_parport_ieee1284_ecp_write_data(struct snx_parport *port, +> +const void *buffer, size_t len, int flags)
+> +{
+> + return 0;
+> +}
+> +
+> +size_t sunix_parport_ieee1284_ecp_read_data(struct snx_parport *port, +> +void *buffer, size_t len, int flags)
+> +{
+> + return 0;
+> +}
+> +
+> +size_t sunix_parport_ieee1284_ecp_write_addr(struct snx_parport *port, +> +const void *buffer, size_t len, int flags)
+> +{
+> + return 0;
+> +}
+
+Why are these all no-ops ?
+
+i will fix it.
+
+> diff --git a/mfd/sunix/snx_lp.c b/mfd/sunix/snx_lp.c
+> new file mode 100644
+> index 00000000..f2478447
+> --- /dev/null
+> +++ b/mfd/sunix/snx_lp.c
+
+<snip>
+
+> +#undef SNX_LP_STATS
+
+what is this ?
+
+i will fix it.
+
+> +static int SNX_PAL_MAJOR;
+> +
+> +#define SNX_LP_NO SNX_PAR_TOTAL_MAX
+> +#undef SNX_CONFIG_LP_CONSOLE
+
+and this ?
+
+i will fix it.
+
+> +#ifdef SNX_CONFIG_PARPORT_1284
+
+dont do #define in the middle of the code.
+
+i will fix it.
+
+> +static const struct file_operations snx_lp_fops = {

+> +     .owner          = THIS_MODULE,
+> +     .write          = snx_lp_write,
+> +     .open           = snx_lp_open,
+> +     .release        = snx_lp_release,

+> +#ifdef SNX_CONFIG_PARPORT_1284
+> + .read = snx_lp_read,
+> +#endif
+> +};
+
+dont reimplement existing standard functionality your own weird way. +use the partport subsystem. see: Documentation/parport-lowlevel.txt +
+i will fix it.
+
+> +static struct snx_parport_driver snx_lp_driver = {

+> +     .name = "lx",
+> +     .attach = snx_lp_attach,
+> +     .detach = snx_lp_detach,

+> +};
+
+yet another case of duplication of some standard struct and hard- +typecasting ? use struct parport_driver here.
+
+i will use standard struct(struct lp_driver) , about struct snx_parport driver, +i will keep current format , because add a list for store device informations. +
+> + SNX_PAL_MAJOR = register_chrdev(0, "lx", &snx_lp_fops);
+
+dont register your own chardev - use the parport subsystem.
+
+i will fix it.
+
+> diff --git a/mfd/sunix/snx_parallel.c b/mfd/sunix/snx_parallel.c +> new file mode 100644
+> index 00000000..461ea4cc
+> --- /dev/null
+
+<snip>
+
+> +struct snx_parport *sunix_parport_pc_probe_port(struct sunix_par_port *priv) +> +{

+> +     struct snx_parport_ops *ops = NULL;
+> +     struct snx_parport *p = NULL;
+> +     struct resource *base_res;
+> +     struct resource *ecr_res = NULL;

+> +
+> + if (!priv)
+> + goto out1;
+> +
+> + ops = kmalloc(sizeof(struct snx_parport_ops), GFP_KERNEL);
+
+why not kzmalloc ?
+
+i will fix it.
+
+> diff --git a/mfd/sunix/snx_ppdev.c b/mfd/sunix/snx_ppdev.c +> new file mode 100644
+> index 00000000..9482ed9f
+> --- /dev/null
+> +++ b/mfd/sunix/snx_ppdev.c
+
+<snip>
+
+> +static const struct file_operations snx_pp_fops = {

+> +     .owner          = THIS_MODULE,
+> +     .llseek         = no_llseek,
+> +     .read           = snx_pp_read,
+> +     .write          = snx_pp_write,
+> +     .poll           = snx_pp_poll,
+> +     .unlocked_ioctl = snx_dump_par_ioctl,
+> +
+> +     .open           = snx_pp_open,
+> +     .release        = snx_pp_release,

+> +};
+
+don't reimplement existing standard functionality - use the parport +subsystem.
+
+i will fix it.
+
+> diff --git a/mfd/sunix/snx_ppdev.h b/mfd/sunix/snx_ppdev.h
+> new file mode 100644
+> index 00000000..0dfec064
+> --- /dev/null
+> +++ b/mfd/sunix/snx_ppdev.h
+> @@ -0,0 +1,15 @@
+> +/* SPDX-License-Identifier: GPL-2.0 */
+> +#include "snx_common.h"
+> +
+> +#define SNX_PP_IOCTL 'p'
+> +
+> +#define SNX_PP_FASTWRITE (1<<2)
+> +#define SNX_PP_FASTREAD (1<<3)
+> +#define SNX_PP_W91284PIC (1<<4)
+
+use the BIT() macro
+
+i will use BIT() macro for bits definition. (DONE)
+
+> diff --git a/mfd/sunix/snx_share.c b/mfd/sunix/snx_share.c
+> new file mode 100644
+> index 00000000..ba6f86a2
+> --- /dev/null
+> +++ b/mfd/sunix/snx_share.c
+> @@ -0,0 +1,629 @@
+> +// SPDX-License-Identifier: GPL-2.0
+> +#include "snx_common.h"
+> +#define SNX_PARPORT_DEFAULT_TIMESLICE (HZ/5)
+> +
+> +unsigned long sunix_parport_default_timeslice = SNX_PARPORT_DEFAULT_TIMESLICE; +> +int sunix_parport_default_spintime = DEFAULT_SPIN_TIME;
+> +
+> +static LIST_HEAD(snx_portlist);
+> +static DEFINE_SPINLOCK(snx_full_list_lock);
+> +
+> +static DEFINE_SPINLOCK(snx_parportlist_lock);
+> +
+> +static LIST_HEAD(snx_all_ports);
+> +static LIST_HEAD(snx_drivers);
+> +
+> +static DEFINE_SEMAPHORE(snx_registration_lock);
+> +
+> +static void sunix_dead_write_lines(
+> +struct snx_parport *p, unsigned char b)
+> +{}
+> +static unsigned char sunix_dead_read_lines(
+> +struct snx_parport *p)
+> +{ return 0; }
+> +static unsigned char sunix_dead_frob_lines(
+> +struct snx_parport *p, unsigned char b, unsigned char c)
+> +{ return 0; }
+> +static void sunix_dead_onearg(struct snx_parport *p)
+> +{}
+> +static void sunix_dead_initstate(
+> +struct snx_pardevice *d, struct snx_parport_state *s)
+> +{}
+> +static void sunix_dead_state(
+> +struct snx_parport *p, struct snx_parport_state *s)
+> +{}
+> +static size_t sunix_dead_write(
+> +struct snx_parport *p, const void *b, size_t l, int f)
+> +{ return 0; }
+> +static size_t sunix_dead_read(
+> +struct snx_parport *p, void *b, size_t l, int f)
+> +{ return 0; }
+> +
+> +

+> +static struct snx_parport_ops        sunix_dead_ops = {
+> +     .write_data                     = sunix_dead_write_lines,
+> +     .read_data                      = sunix_dead_read_lines,
+> +     .write_control          = sunix_dead_write_lines,
+> +     .read_control           = sunix_dead_read_lines,
+> +     .frob_control           = sunix_dead_frob_lines,
+> +     .read_status            = sunix_dead_read_lines,
+> +     .enable_irq                     = sunix_dead_onearg,
+> +     .disable_irq            = sunix_dead_onearg,
+> +     .data_forward           = sunix_dead_onearg,
+> +     .data_reverse           = sunix_dead_onearg,
+> +     .init_state                     = sunix_dead_initstate,
+> +     .save_state                     = sunix_dead_state,
+> +     .restore_state          = sunix_dead_state,
+> +     .epp_write_data         = sunix_dead_write,
+> +     .epp_read_data          = sunix_dead_read,
+> +     .epp_write_addr         = sunix_dead_write,
+> +     .epp_read_addr          = sunix_dead_read,
+> +     .ecp_write_data         = sunix_dead_write,
+> +     .ecp_read_data          = sunix_dead_read,
+> +     .ecp_write_addr         = sunix_dead_write,
+> +     .compat_write_data      = sunix_dead_write,
+> +     .nibble_read_data       = sunix_dead_read,
+> +     .byte_read_data         = sunix_dead_read,
+> +     .owner                          = NULL,

+> +};
+
+
+don't reimplement existing standard functionality. use the parport +subsystem.
+
+can i drop it ? it seems that it has no effect when port gone away. +
+--mtx

Signature · 1 line
2.17.1
#1 lore raw
Quoted 2 lines · Morris Ku

Hi,
Thanks for review, my replies are inline:

I literally have no idea what this is!

It's a reply to a diff of a reply to a patch AFAICS!

I think the best I can do is point you to:

Documentation/process/submitting-patches.rst

Quoted 258 lines

---

(quoted diff) +255 -0
@@ -0,0 +1,255 @@
+
+On 19.03.19 13:08, Morris Ku wrote:
+
+> diff --git a/mfd/sunix/snx_ieee1284_ops.c b/mfd/sunix/snx_ieee1284_ops.c
+> new file mode 100644
+> index 00000000..2dac03fd
+> --- /dev/null
+
+<snip>
+
+> +size_t sunix_parport_ieee1284_read_nibble(struct snx_parport *port,
+> +void *buffer, size_t len, int flags)
+> +{
+> +     return 0;
+> +}
+> +
+> +size_t sunix_parport_ieee1284_read_byte(struct snx_parport *port,
+> +void *buffer, size_t len, int flags)
+> +{
+> +     return 0;
+> +}
+> +
+> +size_t sunix_parport_ieee1284_ecp_write_data(struct snx_parport *port,
+> +const void *buffer, size_t len, int flags)
+> +{
+> +     return 0;
+> +}
+> +
+> +size_t sunix_parport_ieee1284_ecp_read_data(struct snx_parport *port,
+> +void *buffer, size_t len, int flags)
+> +{
+> +     return 0;
+> +}
+> +
+> +size_t sunix_parport_ieee1284_ecp_write_addr(struct snx_parport *port,
+> +const void *buffer, size_t len, int flags)
+> +{
+> +     return 0;
+> +}
+
+Why are these all no-ops ?
+
+i will fix it.
+
+> diff --git a/mfd/sunix/snx_lp.c b/mfd/sunix/snx_lp.c
+> new file mode 100644
+> index 00000000..f2478447
+> --- /dev/null
+> +++ b/mfd/sunix/snx_lp.c
+
+<snip>
+
+> +#undef SNX_LP_STATS
+
+what is this ?
+
+i will fix it.
+
+> +static int SNX_PAL_MAJOR;
+> +
+> +#define SNX_LP_NO SNX_PAR_TOTAL_MAX
+> +#undef SNX_CONFIG_LP_CONSOLE
+
+and this ?
+
+i will fix it.
+
+> +#ifdef SNX_CONFIG_PARPORT_1284
+
+dont do #define in the middle of the code.
+
+i will fix it.
+
+> +static const struct file_operations snx_lp_fops = {
+> +     .owner          = THIS_MODULE,
+> +     .write          = snx_lp_write,
+> +     .open           = snx_lp_open,
+> +     .release        = snx_lp_release,
+> +#ifdef SNX_CONFIG_PARPORT_1284
+> +     .read           = snx_lp_read,
+> +#endif
+> +};
+
+dont reimplement existing standard functionality your own weird way.
+use the partport subsystem. see: Documentation/parport-lowlevel.txt
+
+i will fix it.
+
+> +static struct snx_parport_driver snx_lp_driver = {
+> +     .name = "lx",
+> +     .attach = snx_lp_attach,
+> +     .detach = snx_lp_detach,
+> +};
+
+yet another case of duplication of some standard struct and hard-
+typecasting ? use struct parport_driver here.
+
+i will use standard struct(struct lp_driver) , about struct snx_parport driver,
+i will keep current format , because add a list for store device informations.    
+
+> +     SNX_PAL_MAJOR = register_chrdev(0, "lx", &snx_lp_fops);
+
+dont register your own chardev - use the parport subsystem.
+
+i will fix it.
+
+> diff --git a/mfd/sunix/snx_parallel.c b/mfd/sunix/snx_parallel.c
+> new file mode 100644
+> index 00000000..461ea4cc
+> --- /dev/null
+
+<snip>
+
+> +struct snx_parport *sunix_parport_pc_probe_port(struct sunix_par_port *priv)
+> +{
+> +     struct snx_parport_ops *ops = NULL;
+> +     struct snx_parport *p = NULL;
+> +     struct resource *base_res;
+> +     struct resource *ecr_res = NULL;
+> +
+> +     if (!priv)
+> +             goto out1;
+> +
+> +     ops = kmalloc(sizeof(struct snx_parport_ops), GFP_KERNEL);
+
+why not kzmalloc ?
+
+i will fix it.
+
+> diff --git a/mfd/sunix/snx_ppdev.c b/mfd/sunix/snx_ppdev.c
+> new file mode 100644
+> index 00000000..9482ed9f
+> --- /dev/null
+> +++ b/mfd/sunix/snx_ppdev.c
+
+<snip>
+
+> +static const struct file_operations snx_pp_fops = {
+> +     .owner          = THIS_MODULE,
+> +     .llseek         = no_llseek,
+> +     .read           = snx_pp_read,
+> +     .write          = snx_pp_write,
+> +     .poll           = snx_pp_poll,
+> +     .unlocked_ioctl = snx_dump_par_ioctl,
+> +
+> +     .open           = snx_pp_open,
+> +     .release        = snx_pp_release,
+> +};
+
+don't reimplement existing standard functionality - use the parport
+subsystem.
+
+i will fix it. 
+
+> diff --git a/mfd/sunix/snx_ppdev.h b/mfd/sunix/snx_ppdev.h
+> new file mode 100644
+> index 00000000..0dfec064
+> --- /dev/null
+> +++ b/mfd/sunix/snx_ppdev.h
+> @@ -0,0 +1,15 @@
+> +/* SPDX-License-Identifier: GPL-2.0 */
+> +#include "snx_common.h"
+> +
+> +#define SNX_PP_IOCTL 'p'
+> +
+> +#define SNX_PP_FASTWRITE     (1<<2)
+> +#define SNX_PP_FASTREAD              (1<<3)
+> +#define SNX_PP_W91284PIC     (1<<4)
+
+use the BIT() macro
+
+i will use BIT() macro for bits definition. (DONE)
+
+> diff --git a/mfd/sunix/snx_share.c b/mfd/sunix/snx_share.c
+> new file mode 100644
+> index 00000000..ba6f86a2
+> --- /dev/null
+> +++ b/mfd/sunix/snx_share.c
+> @@ -0,0 +1,629 @@
+> +// SPDX-License-Identifier: GPL-2.0
+> +#include "snx_common.h"
+> +#define SNX_PARPORT_DEFAULT_TIMESLICE        (HZ/5)
+> +
+> +unsigned long sunix_parport_default_timeslice = SNX_PARPORT_DEFAULT_TIMESLICE;
+> +int sunix_parport_default_spintime = DEFAULT_SPIN_TIME;
+> +
+> +static LIST_HEAD(snx_portlist);
+> +static DEFINE_SPINLOCK(snx_full_list_lock);
+> +
+> +static DEFINE_SPINLOCK(snx_parportlist_lock);
+> +
+> +static LIST_HEAD(snx_all_ports);
+> +static LIST_HEAD(snx_drivers);
+> +
+> +static DEFINE_SEMAPHORE(snx_registration_lock);
+> +
+> +static void sunix_dead_write_lines(
+> +struct snx_parport *p, unsigned char b)
+> +{}
+> +static unsigned char sunix_dead_read_lines(
+> +struct snx_parport *p)
+> +{ return 0; }
+> +static unsigned char sunix_dead_frob_lines(
+> +struct snx_parport *p, unsigned char b, unsigned char c)
+> +{ return 0; }
+> +static void sunix_dead_onearg(struct snx_parport *p)
+> +{}
+> +static void sunix_dead_initstate(
+> +struct snx_pardevice *d, struct snx_parport_state *s)
+> +{}
+> +static void sunix_dead_state(
+> +struct snx_parport *p, struct snx_parport_state *s)
+> +{}
+> +static size_t sunix_dead_write(
+> +struct snx_parport *p, const void *b, size_t l, int f)
+> +{ return 0; }
+> +static size_t sunix_dead_read(
+> +struct snx_parport *p, void *b, size_t l, int f)
+> +{ return 0; }
+> +
+> +
+> +static struct snx_parport_ops        sunix_dead_ops = {
+> +     .write_data                     = sunix_dead_write_lines,
+> +     .read_data                      = sunix_dead_read_lines,
+> +     .write_control          = sunix_dead_write_lines,
+> +     .read_control           = sunix_dead_read_lines,
+> +     .frob_control           = sunix_dead_frob_lines,
+> +     .read_status            = sunix_dead_read_lines,
+> +     .enable_irq                     = sunix_dead_onearg,
+> +     .disable_irq            = sunix_dead_onearg,
+> +     .data_forward           = sunix_dead_onearg,
+> +     .data_reverse           = sunix_dead_onearg,
+> +     .init_state                     = sunix_dead_initstate,
+> +     .save_state                     = sunix_dead_state,
+> +     .restore_state          = sunix_dead_state,
+> +     .epp_write_data         = sunix_dead_write,
+> +     .epp_read_data          = sunix_dead_read,
+> +     .epp_write_addr         = sunix_dead_write,
+> +     .epp_read_addr          = sunix_dead_read,
+> +     .ecp_write_data         = sunix_dead_write,
+> +     .ecp_read_data          = sunix_dead_read,
+> +     .ecp_write_addr         = sunix_dead_write,
+> +     .compat_write_data      = sunix_dead_write,
+> +     .nibble_read_data       = sunix_dead_read,
+> +     .byte_read_data         = sunix_dead_read,
+> +     .owner                          = NULL,
+> +};
+
+
+don't reimplement existing standard functionality. use the parport
+subsystem.
+
+can i drop it ? it seems that it has no effect when port gone away.
+
+--mtx
Signature · 4 lines
Lee Jones [李琼斯]
Linaro Services Technical Lead
Linaro.org │ Open source software for ARM SoCs
Follow Linaro: Facebook | Twitter | Blog
#2 lore raw
In reply to Morris Ku
Quoted 2 lines · Morris Ku

Hi,
Thanks for review, my replies are inline:

https://www.netmeister.org/news/learn2quote.html

Quoted 10 lines

+> +static struct snx_parport_driver snx_lp_driver = {

+> +     .name = "lx",
+> +     .attach = snx_lp_attach,
+> +     .detach = snx_lp_detach,

+> +};

(quoted diff) +4 -0
+
+yet another case of duplication of some standard struct and hard-
+typecasting ? use struct parport_driver here.
+

i will use standard struct(struct lp_driver) ,

you mean struct parport_driver ?

Quoted 2 lines

about struct snx_parport driver,
i will keep current format , because add a list for store device informations.

No, your current approach breaks heavily (eg. as soon ans something in struct parport_driver changes). If you really need to extend it, do it by nested structs (same like struct parport_driver nests struct device_driver), and use the proper macros (eg. container_of(), etc) for typecasting.

Keep in mind: kernel-internal structures are NOT fixed, they can change anytime - this is a fundamental design decision.

Review

Quoted 4 lines
(quoted diff) +4 -0
+don't reimplement existing standard functionality. use the parport
+subsystem.
+
+can i drop it ? it seems that it has no effect when port gone away.

If it's not needed, remove it. We don't like dead code.

--mtx

Signature · 3 lines
Enrico Weigelt, metux IT consult
Free software and Linux embedded engineering
[email protected] -- +49-151-27565287