Question about the use of realloc vs. memory pools

[alejandro-perez]

Hello,

since I'm trying to get good knowledge of the GSS attribute exportation code, I've come across the use of realloc, which is surprising to me given that memory is usually allocated by means of apr_* functions and pools. Is not realloc using plain "malloc" underneath? If so, shouldn't the code use the pool?

I've seen two uses of realloc corresponding to the GSS attribute handling.

1. https://github.com/modauthgssapi/mod_auth_gssapi/blob/master/src/mod_auth_gssapi.c#L1631-L1644
   Here it seems that at least a cleanup fnc is registered, so memory should be freed.

2. https://github.com/modauthgssapi/mod_auth_gssapi/blob/master/src/environ.c#L43-L51
   In this case, I don't see any cleanup function. How is this memory freed?

[simo5]

On Thu, 2017-02-09 at 10:29 -0800, Alejandro Perez wrote: I've seen two uses of realloc corresponding to the GSS attribute handling. https://github.com/modauthgssapi/mod_auth_gssapi/blob/master/src/mod_ auth_gssapi.c#L1631-L1644 Here it seems that at least a cleanup fnc is registered, so memory should be freed.

I think that this was done with realloc because there is no function to free data with pools (as the assumption is that everything is freed once you destroy the pool). That's why there is a cleanup function.

https://github.com/modauthgssapi/mod_auth_gssapi/blob/master/src/envi ron.c#L43-L51 In this case, I don't see any cleanup function. How is this memory freed?

This is probably a copy/paste bug, and the pool should be used. Simo.

…

      {"api_version":"1.0","publisher":{"api_key":"05dde50f1d1a384dd78767c5 5493e4bb","name":"GitHub"},"entity":{"external_key":"github/modauthgs sapi/mod_auth_gssapi","title":"modauthgssapi/mod_auth_gssapi","subtit le":"GitHub repository","main_image_url":"https://cloud.githubusercon tent.com/assets/143418/17495839/a5054eac-5d88-11e6-95fc- 7290892c7bb5.png","avatar_image_url":"https://cloud.githubusercontent .com/assets/143418/15842166/7c72db34-2c0b-11e6-9aed- b52498112777.png","action":{"name":"Open in GitHub","url":"https://github.com/modauthgssapi/mod_auth_gssapi"}},"u pdates":{"snippets":[{"icon":"DESCRIPTION","message":"Question about the use of realloc vs. memory pools (#126)"}],"action":{"name":"View Issue","url":"https://github.com/modauthgssapi/mod_auth_gssapi/issues /126"}}}

[alejandro-perez]

On Thu, 2017-02-09 at 10:29 -0800, Alejandro Perez wrote: > I've seen two uses of realloc corresponding to the GSS attribute > handling. > > > > https://github.com/modauthgssapi/mod_auth_gssapi/blob/master/src/mod_ > auth_gssapi.c#L1631-L1644 > > Here it seems that at least a cleanup fnc is registered, so memory > should be freed. I think that this was done with realloc because there is no function to free data with pools (as the assumption is that everything is freed once you destroy the pool). That's why there is a cleanup function.

Makes sense.

> > https://github.com/modauthgssapi/mod_auth_gssapi/blob/master/src/envi > ron.c#L43-L51 > > In this case, I don't see any cleanup function. How is this memory > freed? > > > This is probably a copy/paste bug, and the pool should be used. Simo.

I'm not sure about that. It uses realloc as in the example above to assure array grows in 16 modules. Maybe what it is missing is the cleanup function? I can take care of that, if you want.

[simo5]

On Fri, 2017-02-10 at 03:16 -0800, Alejandro Perez wrote: I'm not sure about that. It uses realloc as in the example above to  assure array grows in 16 modules. Maybe what it is missing is the  cleanup function?

My reasoning was that mc has a well defined lifetime and it is probably ok to "leak" some memory in it as it gets cleaned up relatively quickly. However a cleanup function would work as well.

I can take care of that, if you want.

Sure, thanks!

[alejandro-perez]

Fixed in #127