Tarantool development patches archive
 help / color / mirror / Atom feed
From: Cyrill Gorcunov <gorcunov@gmail.com>
To: Vladislav Shpilevoy <v.shpilevoy@tarantool.org>
Cc: tml <tarantool-patches@dev.tarantool.org>
Subject: Re: [Tarantool-patches] [PATCH v8 1/4] box/func: factor out c function entry structure
Date: Fri, 30 Oct 2020 12:51:57 +0300	[thread overview]
Message-ID: <20201030095157.GD198833@grain> (raw)
In-Reply-To: <8706dd6e-0b8d-f3f1-2344-f1b24906eecd@tarantool.org>

On Thu, Oct 29, 2020 at 11:15:51PM +0100, Vladislav Shpilevoy wrote:
...
> > +int
> > +module_sym_load(struct module_sym *mod_sym)
> > +{
> > +	assert(mod_sym->addr == NULL);
> > +
> > +	struct func_name name;
> > +	func_split_name(mod_sym->name, &name);
> > +
> > +	struct module *module = module_cache_find(name.package,
> > +						  name.package_end);
> > +	if (module == NULL) {
> > +		/* Try to find loaded module in the cache */
> > +		module = module_load(name.package, name.package_end);
> > +		if (module == NULL)
> > +			return -1;
> > +		if (module_cache_put(module)) {
> 
> 1. Please, use explicit != 0.
> https://github.com/tarantool/tarantool/wiki/Code-review-procedure#code-style

OK

> 
> > +			module_delete(module);
> > +			return -1;
> > +		}
> > +	}
> > +
> > +	mod_sym->addr = module_sym(module, name.sym);
> > +	if (mod_sym->addr == NULL)
> > +		return -1;
> 
> 2. If the module was loaded first time here, it is not unloaded in case of
> an error in this place.

Just like it was before the patch. The patch simply factor outs the old
code. It doesn't improve it (because, lets be honest this is a min problem
for module management -- we've to check for module symbol not at moment
of calling it but rather at the moment when we load a function). That
said the issue with module management is known and I think we need to
rework modules code more deeply, but not in this series. In the
series it remains exactly as it was.

> >  
> > +/**
> > + * Callable symbol bound to a module.
> > + */
> > +struct module_sym {
> > +	/**
> > +	 * Anchor for module membership.
> > +	 */
> > +	struct rlist item;
> > +	/**
> > +	 * For C functions, address of the function.
> > +	 */
> > +	box_function_f addr;
> > +	/**
> > +	 * Each stored function keeps a handle to the
> > +	 * dynamic library for the C callback.
> > +	 */
> 
> 3. Can't parse the comment. What is the 'C callback'?
> And why is this function stored? After you extracted it
> from struct func_c, it is not related to _func space, and
> is not stored.

It is stored in memory. The C callback is the function we
call, so we keep a pointer to a module. If shuch comment
confuses lets change it to something like "Each function
keeps a handle to the module where the symbol relies"?

  reply	other threads:[~2020-10-30  9:52 UTC|newest]

Thread overview: 22+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-10-14 13:35 [Tarantool-patches] [PATCH v8 0/4] box/cbox: implement cfunc Lua module Cyrill Gorcunov
2020-10-14 13:35 ` [Tarantool-patches] [PATCH v8 1/4] box/func: factor out c function entry structure Cyrill Gorcunov
2020-10-29 22:15   ` Vladislav Shpilevoy
2020-10-30  9:51     ` Cyrill Gorcunov [this message]
2020-10-31  0:13       ` Vladislav Shpilevoy
2020-10-31 15:27         ` Cyrill Gorcunov
2020-10-14 13:35 ` [Tarantool-patches] [PATCH v8 2/4] module_cache: move module handling into own subsystem Cyrill Gorcunov
2020-10-29 22:15   ` Vladislav Shpilevoy
2020-10-30 10:15     ` Cyrill Gorcunov
2020-10-31  0:15       ` Vladislav Shpilevoy
2020-10-31 15:29         ` Cyrill Gorcunov
2020-10-14 13:35 ` [Tarantool-patches] [PATCH v8 3/4] box/cbox: implement cbox Lua module Cyrill Gorcunov
2020-10-29 22:15   ` Vladislav Shpilevoy
2020-10-30 12:51     ` Cyrill Gorcunov
2020-10-31  0:21       ` Vladislav Shpilevoy
2020-10-31 21:59         ` Cyrill Gorcunov
2020-11-01  8:26           ` Cyrill Gorcunov
2020-11-02 22:25           ` Vladislav Shpilevoy
2020-11-03  7:26             ` Cyrill Gorcunov
2020-11-12 22:54     ` Vladislav Shpilevoy
2020-11-13 18:30       ` Cyrill Gorcunov
2020-10-14 13:35 ` [Tarantool-patches] [PATCH v8 4/4] test: box/cfunc -- add simple module test Cyrill Gorcunov

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20201030095157.GD198833@grain \
    --to=gorcunov@gmail.com \
    --cc=tarantool-patches@dev.tarantool.org \
    --cc=v.shpilevoy@tarantool.org \
    --subject='Re: [Tarantool-patches] [PATCH v8 1/4] box/func: factor out c function entry structure' \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox